Tab 2 - Events | Board of Educationlaschoolboard.org/sites/default/files/10-08-19CIATab2...2019/10/08  · by Gov. Jeb Bush. All North Carolina public schools, including charters,

Tab 2

Board Member Resolution for Discussion

Ms. Goldberg – Ensuring Support for Students at All Los Angeles Unified School District

Schools (Res-011-19/20) (Noticed September 3, 2019)

Whereas, The value of a public school cannot be quantified in a single, summative rating, which

can shame, penalize, or stigmatize schools, education professionals, students, and entire

communities;

Whereas, School ratings promote unhealthy competition between schools, exacerbate community

antagonisms by producing artificial “winners” and “losers”, and penalize schools that serve

socio-economically disadvantaged student populations;

Whereas, It is well-understood that the academic achievement gap continues to be perpetuated by

socio-economic and socio-emotional factors;

Whereas, School ranking systems that prioritize student performance on standardized tests fail to

adequately take into account socio-economic and socio-emotional factors that influence school

needs and are reflected in the academic achievement levels of students;

Whereas, Research indicates that online school performance achievement ratings exacerbate

economic and social segregation in the United States, accelerating divergences in housing values,

income distributions, education levels, and the racial and ethnic composition of communities;

Whereas, On April 3, 2018, the Governing Board of the Los Angeles Unified School District

adopted Achieving Excellence for All: Establishing a Framework for Continuous Improvement

(Res-036-17/18), which called for the development of a School Performance Framework (SPF)

that would enable the District “to identify and track the overall performance of each school

annually” and “evaluate all schools with a uniform set of measures”;

Whereas, The proposed SPF presented to the Board on November 13, 2018, heavily relies on

Value-Added “Growth” modeling, a technique subject to variability and error, which is therefore

inappropriate for high-stakes evaluations resulting in rankings, categorization, or absolute

comparisons;

Whereas, The proposed SPF seeks included the requirement to distill the natural and inherent -

complexity of a school community into a single, summative rating;

Whereas, Identifying schools and students meeting or not meeting proficiency level benchmarks

to inform the provision of support and resources is a task already conducted by District

administrators, principals, and teachers on an almost daily basis; and

Whereas, Rather than having to implement a summative school ranking system, the

Superintendent and Local District Superintendents should be permitted to focus on continuous improvement using multiple measures such as student growth, responsible reclassification and school climate to addressing the academic achievement gap experienced by students from

socioeconomically disadvantaged communities lacking financial, political, and social capital;

now, therefore, be it

Resolved, That the Los Angeles Unified School District suspend implementation of the SPF and

any launching or utilization of the SPF—including any use of stars, scores, or any other rating

system—in or on any District platforms until a more in-depth analysis of the SPF’s metrics is

conducted by the members of the Governing Board of the Los Angeles Unified School District

and/or an ad-hoc committee to understand the implications and impact of the SPF on District

schools and their residing neighborhoods;

Resolved further, That the District further the goal of giving students, parents, school employees,

and the general public more information about District schools by including important and wide-

ranging types of data from the California Department of Education’s California School

Dashboard on its website in an accessible fashion;

Resolved further, That the District will provide parents with a guide and workshops on how to

utilize the California School Dashboard;

Resolved further, That the District collaborate with the California Department of Education to

improve the California School Dashboard based on school community and parent feedback;

Resolved further, That Local District personnel work closely with the administrators and

teachers of schools with large numbers of students experiencing difficulties meeting or

exceeding state standards to help increase the number of students meeting and exceeding state

standards at these schools; and, be it finally

Resolved further, The funds designated to implement the SPF be equitably distributed according

to need between: (i) Local Area Districts tasked with identifying schools that serve high

concentrations of high-needs students experiencing difficulty performing academically; and (ii)

supports that can be provided to increase academic performance among those students; and, be it

finally

Resolved, That the schools, both District and charter, that have similar locations, grade-level

spans and socioeconomic status, convene each spring to reciprocally share “best practices,” in a

collaborative all-schools meeting facilitated by Local Area Districts dedicated to encouraging

cross-school communication that stimulates successful innovation and learning for all our

students.

Documents

Submitted by Dr.

McKenna

Document 1

See how closely Ohio school report card grades trend with

district income On Tuesday, the Cleveland Plain Dealer published a stunning analysis, by the newspaper’s data

analyst Rich Exner, of the school district grades awarded by the state of Ohio on the state report

cards released last week. The new report cards are based on data from the 2018-2019 school

year.

CLEVELAND, Ohio - The latest set of Ohio school report cards not only provided a scorecard for each

district statewide - they once again drove home the point that wealthier districts do better on such

reports.

For example, incomes in the "A" districts were three times higher than those in the "F" districts, and the

child poverty rate was 13 times higher in the worst performing districts, cleveland.com found.

To get an idea of how closely report card grades from the Ohio Department of Education follow

demographic factors, cleveland.com compared those grades to U.S. Census Bureau community data for

household income, child poverty and the education level of the adults.

In nearly every key report card category, the trends followed census data closely.

For example, taking the median household income for each district, the average among those getting

"A" overall grades was $95,423. It was $65,307 for B-graded districts, $54,058 for C-graded districts,

$44,428 for D-graded districts and $32,658 for F-graded districts.

In the A districts, 58.5% of the adults age 25 and older have at least a bachelor's degree. That share

drops to 17.1% for D-graded districts and 16.3% for F-graded districts.

There are outliers, of course. They are highlighted at this link, ranking each Ohio school district while

factoring in income. But overall, the trends hold true.

The descriptions for each category shown below are from the Ohio Department of Education. The

census data cleveland.com used to make the calculations is from the 2017 American Community Survey.

Achievement component grade

The Achievement component represents whether student performance on state tests met established

thresholds and how well students performed on tests overall. A new indicator measures chronic

absenteeism.

Graduation rate component

The Graduation rate component looks at the percent of students who are successfully finishing high

school with a diploma in four or five years.

Progress component

The Progress component looks closely at the growth that all students are making based on their past

performances.

Gap closing component

The Gap closing component shows how well schools are meeting the performance expectations for our

most vulnerable students in English language arts, math, graduation and English language proficiency.

Improving at-risk K-3 readers component

This component looks at how successful the school is at improving at-risk K-3 readers.

Prepared for success component

Whether training in a technical field or preparing for work or college, the prepared for success

component looks at how well prepared Ohio’s students are for all future opportunities.

Rich Exner, data analysis editor for cleveland.com, writes about numbers on a variety of topics. Follow

on Twitter @RichExner. Go to cleveland.com/datacentral for other data-related stories and analysis.

Related report card information

Searchable database: Find report card details for each Ohio school district, the buildings within each

district and charter schools.

Ohio school districts ranked 1 to 608, based on the report card's Performance Index.

Cleveland, Euclid and Lorain avoid Fs on state report cards, while four districts earn new As

Ohio Department of Education

Document 2

North Carolina: School Grades Measure Income Not School Quality

by dianeravitch

Public Schools First in North Carolina posted an analysis of the grades given to schools by the state,

based mostly on test scores. Not surprisingly, the school grades measured income, not school quality,

since standardized tests measure income.

School Performance Grades

NC SCHOOL PERFORMANCE GRADES 2018-19 BY

INCOME LEVEL

Low income range

Source: N&O analysis of Public Instruction data

School performance grades started in 2013-14 modeled after a program in Florida started

by Gov. Jeb Bush. All North Carolina public schools, including charters, have received A-

F performance grades since 2013.

Critics of a single school measurement believe that grades:

• Do not reflect the learning in our schools

• Undervalue student growth and other important measures of school quality

• Could result in more attention to borderline students while underserving the lowest and highest

performing students

• Are often used by privatization advocates to support school choice measures and state

takeovers of schools, removing these schools from local control and community input.

• Will have negative economic impacts on a community (lower home values/sales)

• Do not come with resources/financial support to improve grades

How did North Carolina's Schools do This Year? Results show that these grades continue to be

closely correlated with a student’s family income level.

• Schools with greater poverty earned fewer A/A+NG's and B's and earned more C's, D's, and F's

than schools with less poverty.

• Of the 21.7 percent of schools receiving a D or F grade, 95 percent were serving high poverty

populations.

• In schools with more than 80 percent low income students, 60 percent received a D or F grade. Less

than one percent of schools with less than 20% low income student populations received a D or F

grade.

• Of schools with high concentrations (41 percent or more) of students who are economically

disadvantaged, 71 percent met or exceeded growth, compared with 79 percent of schools serving

fewer students in poverty.

• For the 2018-19 school year, 73.3 percent of all schools met or exceeded growth expectations, a

slight increase from the previous year.

Document 3

OCTOBER 2019 DENVER, CO

School Performance Framework:

What Does it Measure in 2017? November 1, 2017 / Melinda Pearson / Announcements, Community Issues, Denver News, Kids,

Schools/Education /

Denver Public Schools (DPS) released results of its School Performance Framework (SPF) measure

last month. The SPF provides a color-coded “report card” for each of the nearly 200 schools in DPS.

This year brings a marked increase in the number of schools deemed to be meeting or exceeding

expectations.

The SPF includes student proficiency and academic growth, as well as student and family

engagement and in high school, college and career readiness. Download a pdf of the above chart

here.

Growth is measured by how much a student improves academically on standardized tests over a

year compared to peers who started at the same place. In the SPF, growth is approximately three

times as important as academic proficiency in elementary and middle school and twice the weight of

proficiency in high school. These are the same ratios as used last year.

What this means is that schools that are academically “growing” their students—regardless of their

overall proficiency—can perform better in the SPF ratings than schools whose students are meeting

or exceeding expectations but simply maintaining that proficiency. This reflects the district’s

emphasis on increasing student progress, particularly at historically low-performing schools.

“The SPF should be the first thing you check out to make sure it’s a safe place and kids are

learning…

And then you have to figure out to what degree…

and what are they learning, and that requires a lot more research.”

-–Van Schoales , CEO, A+ Colorado

Local Schools Performance

The charts included here give an overview of how schools in the northeast Denver area performed

on the SPF this year. There are five levels on the rating scale: blue/distinguished, green/meets

expectations, yellow/accredited on watch, orange/accredited on priority watch, and red/accredited on

probation.

The SPF is used by schools and the district to evaluate school performance, and it is also a resource

for families to understand how schools are educating their kids compared to district expectations.

But it can be a challenge to serve both purposes. “We are really trying to provide that information in

… a more graphical way or just a way that our parents and families want that information to be

provided,” said Katherine Beck, senior manager of accountability for DPS’s Accountability Research

and Evaluation division. To that end, DPS provides summary reports and detailed reports on each of

its schools, available here: http://spf.dpsk12.org/en/2017-spf-ratings/.

But Van Schoales, CEO of A+ Colorado, a nonprofit organization that analyzes educational efforts

statewide, says that truly understanding all that goes into the SPF is “so complicated [that] I compare

it to the IRS tax code.” Once you start to delve into the data—and there are hundreds of data points

that go into each school’s rating—it can get overwhelming for families.

According to Schoales, for “good” schools that are in the high green/blue range “the difference really

comes down to what the needs of that family and that child are.” He cautions against putting too

much emphasis on differences of just a few points between schools. But in the range of “low green,

red, yellow and orange, unfortunately, that’s a murky area. It’s hard to know how bad things are,” he

added.

Changes This Year in SPF

Each year, DPS adjusts the formula used to determine SPF results to better align with district goals

and community needs, said Beck. The ambitious 2020 plan emphasizes five specific areas:

increasing quality schools, early literacy, college and career readiness, whole child support, and

closing the opportunity gap.

Academic Gaps Indicator

Included within the proficiency and growth scores are measures of “academic gaps,” which evaluate

the performance of underserved students—such as poor students, English language learners,

students of color, and special education students—compared to their peers. This year, the academic

gaps rating is not only made evident with its own indicator, but schools that have failed to meet

criteria in that area have been penalized. This is in line with DPS’s emphasis on closing opportunity

gaps.

Districtwide, nine schools that would otherwise have achieved a “green/meets expectations” overall

rating were knocked down to “yellow/accredited on watch” based on their academic gaps, according

to Beck. “Even at some of our highest-performing schools, are all of the students in that school

performing at that equally very high level?” asks Beck.

For example, Bromwell Elementary, which has historically been a high green school—and in fact this

year achieved a very high 74.48 percent of points on the SPF—has been downgraded to a “yellow”

rating due to academic gaps between underserved students and their classmates.

Eight schools in the east Denver area—including Denver Discovery School, Hill Campus of Arts and

Sciences and George Washington High School—did not meet expectations for academic gaps, but

none lost a green rating based on the indicator alone. Those equity gaps were factored into their

overall performance, among other markers, and the aggregate scoring resulted in lower ratings.

Schoales believes the district overemphasizes the importance of the gaps between students. “We

think it’s a [bigger] problem to have lower levels of growth and proficiency,” he said. The emphasis

on the gaps over general levels of growth and proficiency, “provides, frankly, a disincentive for

schools to have a diverse population, which we think is a huge problem.”

Early Literacy Measures

Another significant change this year for elementary schools is the inclusion of early literacy

measures. These are tests for students in grades K–3, and Beck says that they were added because

early literacy “wasn’t as well captured in previous SPFs.” She added that the measures used were

“ones that are focused on historically disadvantaged learners—students of color, students on free

and reduced lunch, English language learners, and special education students.”

But according to Schoales, “those [early literacy] tests don’t line up with what the district and the

state have said is standard in terms of kids’ reading and writing.” In fact, he continued, “we’ve found

that there is over a 30 percent difference between results from those tests and state tests/CMAS.

That, in turn, dramatically increases the scores on the SPF.” His organization’s analysis found that

many schools that were rated highly based on early literacy tests—which students can take

repeatedly—show growth and proficiency that are below state benchmarks. “The SPF misrepresents

what a quality school is because of those indicators,” said Schoales.

Future Changes

With DPS’s ambitious goals to ensure that 80 percent of the district’s third-graders are reading at or

above grade level and 80 percent of its students attend a green or blue school by 2020, the district

will be setting the bar higher next year. Beck said that the district is moving to more rigorous SPF

standards for the statewide testing measures in 2018 and then for early literacy measures in 2019,

giving schools plenty of advance notice to rise to the challenge.

Document 4

Apples to Apples: Common School Performance Frameworks as a Tool for Choice and Accountability 1

In most cities today, school choice is the norm, not the exception. According to our recent survey of eight U.S. cities, 55 percent of parents said their child attends a school other than their neighborhood school. Students in cities are likely to have a choice between a wide array of district-run magnet and alternative schools, public charter schools, and private schools. Nationwide, 6 percent of all public school children attend a charter school and 13 percent of children in traditional public schools attend a school their parents chose rather than the one they were assigned to, according to the National Center for Education Statistics.

Cities across the country continue to offer more public school options to their students. The number of charter schools more than doubled in the past decade. And dozens of school districts are now employing a “portfolio strategy,” expanding and diversifying school options while holding all schools, no matter the type, accountable for performance. Families living in the Cleveland Metropolitan School District, for example, can send their children to a charter school, or a district school with a focus on digital art, architectural design, or environmental studies, or a neighborhood school with a traditional curriculum.

In the midst of this dramatically changing face of public education, the need for meaningful, comparable, and understandable measures of school performance has never been more urgent. Urban parents need information to help them make choices among a sometimes dizzying array of options. Schools need to be able to understand how their performance compares to other schools. And government agencies overseeing the city’s entire portfolio of schools need ways to make a wide range of decisions, including which schools should be expanded or replicated, receive targeted support, or be replaced.

An Overview of Common School Performance FrameworksSarah Yatsko and Alice Opalka Center on Reinventing Public Education

Yet within most cities, district schools, charter networks, and independently operated charter schools provide different sets of performance information to different regulators and groups and are held to different performance standards and measures of quality. In Detroit, for instance, public schools are overseen by a dozen different charter oversight agencies, the Detroit Public School District, and the state-run Education Achievement Agency, all with their own methods of rating schools and holding them accountable. In Detroit, St. Louis, and Memphis the situation has become so confusing for families that local nonprofits and advocacy groups have begun publishing their own rating systems to inform parents and, in the case of Detroit, to ensure that low-performing schools are closed.

To address problems like these, some cities are developing new citywide performance systems that (1) provide consistent information to families and schools and (2) provide one tool to allow government agencies to make informed and fair school oversight decisions. These systems are referred to as Common School Performance Frameworks (CSPF) in this report; they also are sometimes called Unified Accountability Systems.

As part of the Center on Reinventing Public Education’s (CRPE) ongoing work to understand and support successful portfolio management and district-charter collaboration, this report is meant to be a resource for city leaders who are considering implementing or refining a CSPF across all schools, both district and charter. CRPE studied the experiences of six cities, including conducting interviews with district and charter leaders in Chicago, Los Angeles, Denver, and New Orleans1 and working directly with leaders in Memphis and Sacramento to look more deeply at the process. See Table 1 for a description of 11 cities that currently, formerly, or will soon use various types of school performance reports. The goal was to document lessons about how these frameworks are developed and implemented to inform other cities.

1 The Louisiana Charter School Performance Compact was developed as a performance accountability tool for all charter schools authorized by the Board of Elementary and Secondary Education. This Charter School Performance Compact applies to over 70 percent of students in New Orleans as of this writing.

Style Guide

Apples to Apples: Common School Performance Frameworks as a Tool for Choice and Accountability 2

To gather the information presented here, CRPE contracted with researchers and practitioners who observed and supported two cities through their efforts to develop CSPFs. This guide includes two important resources:

Lessons from Chicago: Developing a Common School Performance FrameworkThis case study, written by Jessica Sutter of EdPro Consulting, describes Chicago’s efforts in developing its School Quality Rating Policy. It is an illustration of one city’s approach to a CSPF, including how it overcame common hurdles such as district-charter tensions and data limitations.

How to Develop a Common School Performance FrameworkLaura Weeldreyer of UPD Consulting and and David Stewart, founder of Tembo, Inc., prepared this practical how-to guide for cities at or just beyond the decision point to develop a CSPF. To prepare this report, Weeldreyer draws on her two decades of work in education, both as a high-level district administrator and as a consultant to districts and state education agencies. She also spent 10 months facilitating the creation of CSPF in a large city and has compiled insights and advice on how to create a framework with input from well-informed key stakeholders using reliable and available data that track progress toward agreed-upon goals. Stewart’s contributions include some of the more technical aspects of choosing and defining metrics.

These resources are valuable for any “high-choice” city looking for ways to inform parent choice, government oversight, or strategies for portfolio management of the system of schools. They will deeply inform both the technical and political challenges involved in rethinking school accountability and provide specific guidance on how those challenges can be overcome.

They also provide important ideas and lessons to any city or state that is grappling with how to develop rich accountability measures under the Every Student Succeeds Act (ESSA). They offer essential guidance about which metrics cities commonly consider, including how they prioritize them and what political and technical considerations one should give to them.

What Is a Common School Performance Framework?

A CSPF is a tool for measuring the performance of an individual school, using a defined set of metrics that is common to schools across different agencies or governing bodies. CSPFs are shared between district-run schools, charter schools, contracted public schools, and charter networks within the same city or geographic area. The goal is to enable comparisons across and between all public schools in the area, regardless of who runs them.

CSPFs are specific to a region (typically a city), as opposed to a state school rating or accountability system. They allow a city to set their own priorities for how school quality is defined and measured that go beyond any state definitions or tests, and while state rating systems usually rely solely upon standard testing data, CSPFs commonly incorporate multiple metrics, including non-academic measures such as school climate.

Cities often provide families with common data to inform school choice and to provide transparency (for example, information in the Washington, D.C., Equity Reports). A CSPF goes further, providing an evaluative rating that is easy for parents and community members to digest. One important goal, although rarely achieved, is to ensure there are common

consequences where schools in the same ratings categories get the same treatment across sectors. For example, when any public school, district or charter, falls below an agreed upon performance standard, it is provided with support or intervention services, or it is replaced with a school that has been consistently rated as high performing and is ready to expand.

Despite these common goals, cities that have invested in fully developed CSPFs have all taken somewhat different approaches, as shown below.

The purpose of the tool. Cities vary to what extent they attach consequences to the CSPF. Some use the data simply to highlight successful schools, place informal pressure on low-performers, and inform parents. Most, however, use the framework to help officials decide:

s Which schools to target for additional support, intervention, or closure.

s Whether to allow charter schools to replicate or have access to district facilities.

For cities pursuing a portfolio management strategy, CSPFs are a key component to ensuring continuous improvement and equity across the city. Denver, for

Apples to Apples: Common School Performance Frameworks as a Tool for Choice and Accountability 3

example, uses its CSPF to determine where high-quality schools are located, which neighborhoods are in need of better options, which schools should be replicated, which schools are in need of intervention or support, and which schools should be closed.

The types of measures included in the tool. Standardized test scores are almost always included, but cities differ in whether they emphasize proficiency or growth scores or in how they break scores down by different student populations or sub-groups. Most cities, however, also use other measures of quality, such as Advanced Placement enrollment rates, high school exit exams, and measures of college readiness such as SAT test performance. Beyond academics, cities also have decided to incorporate school climate that can be gleaned through measures including student attendance rates; teacher turnover; teacher, student, and/or parent surveys; or rates of use of discipline, including suspensions and expulsions.

Some cities, such as New Orleans and Los Angeles, intentionally use very few measures of performance to make it clear when schools will face state intervention and to ensure that parents can quickly and easily assess how a school is faring. With a large number of charter schools overseen by the state, the Louisiana Department of Education wanted a common way to measure school performance for all charter schools authorized by Louisiana’s Board of Elementary and Secondary Education. Because of the unique nature of the New Orleans school system (a nearly all-charter school system), Louisiana made sure that the autonomy of charter schools was preserved in the process; the state intentionally kept metrics very simple and understandable, and it made sure not to infringe on schools’ individual priorities. Denver, on the other hand, uses roughly 80 different metrics to create a comprehensive tool that allows school operators to evaluate performance using multiple measures yet can be streamlined to provide parents with the information they need to choose the best fit for their child.

The process for gathering stakeholder input and managing the political process. Every city studied for this report gathered input from school staff and others, but the process differed in each city. In Nashville, an alignment of frameworks across sectors evolved with little fanfare. Charter school authorizers, with the support of the National Association of Charter School Authorizers (NACSA), developed a strong performance framework that they then shared with the Metro-Nashville Public Schools. The district saw it as an improvement on what they had been using and decided to use parts of it themselves. In contrast,

those who led Chicago’s CSPF efforts pointed to extensive public engagement, many focus groups, and lots of work to help schools understand and calculate their scores in advance of public release as key to successful implementation. The case study included in this report provides more detail on that process. In Baltimore, as the sole charter school authorizer, the district heavily engaged with the charter sector on the development of a performance framework that would be used to make charter renewal decisions. The end result in 2013 was a framework that was aligned, but not identical to the framework the district had simultaneously developed to hold traditional schools accountable. Yet charter school leaders still publically supported the new charter accountability tool, thankful to have provided input during its development.

Sacramento’s effort to elicit broad input was easily the most comprehensive example studied, with tens of focus groups with schools and community members and with district officials making dozens of home visits to interview families in the city. The framework’s design iterated throughout the process based on evolving feedback. But public engagement alone is not enough. Without the development of a plan for sustained momentum or buy-in from the incoming administration, the framework did not survive a transition of superintendents and other district administrative staff.

Similarly, in Memphis, even after a strong public engagement campaign and months of facilitated meetings between Shelby County Schools and the state’s Achievement School District (ASD), the effort to create a CSPF was unsuccessful. Differences emerged over metrics and their weights and how schools, particularly the lowest-performing ones, would ultimately be represented in the framework. The ASD has since adopted their own aligned—although not identical—accountability frameworks and, as of this writing, continue to work through a broader engagement with the district.

The effort in Memphis illustrates how critical it is for those involved to maintain a high level of commitment throughout the development process. Luckily, a local community organization independently developed the Memphis School Guide for parents, a tool to compare all district and charter schools in the city, after sensing that the CSPF process could be derailed. Though not as robust as a full CSPF given limited publically available data that it draws upon, the Memphis School Guide does fill a gap by providing parents with easy access to public school performance and other information to support parent choice.

Apples to Apples: Common School Performance Frameworks as a Tool for Choice and Accountability 4

The technical decisions made about how to ensure that the tool is analytically sound and understandable. The process by which city leaders develop a performance management tool that will be used for high-stakes decisions such as charter renewal must be able to withstand criticism and challenge. For that reason, many decisions need to be made:

s What data sources to include?

s How to reconcile different data sources across district and charter systems or how to collect new kinds of data?

s How to weight different factors?

s Whether to roll up the different metrics into one summative “rating” or to present independent ratings for specific domains, such as school climate and academic achievement?

Each city studied had different preferences and approaches to these questions. The accompanying guide goes into depth about the tradeoffs involved in different approaches and examples of approaches used.

Avoiding Landmines

Any time performance data are compiled, presented, and especially when used for high stakes decisions, much can go wrong. Charter school leaders can resist being pulled into a system that they believe will limit their autonomies or that will impose a new set of performance standards that they did not originally agree on. District schools may fear that new consequences will be attached to the framework on top of current expectations. A variety of community leaders, union representatives, and others all may have very strongly held views about the “correct” measures of school quality.

CRPE’s research across all six cities offers some critical high-level lessons for avoiding landmines and making sure CSPFs can be effective long-term tools for school improvement throughout a city.

1. Carefully tend to politics and engagement to build long-term sustainability. A CSPF can be rushed in implementation or forced on schools through dictates, (i.e., a district charter authorizer can unilaterally develop and enforce a CSPF) but there may be a significant price to pay in political backlash or simply a lack of commitment when those who pushed the idea move out of the city.

Those implementing a CSPF should pay close attention to the political context in their city and consider which individuals and organizations need to be involved in the process to improve the metrics and to ensure that there is a long-term commitment to its use. Having firm commitment from the superintendent and key charter leaders is essential and should be cultivated and managed appropriately.

Chicago CSPF leaders did extensive outreach throughout the city and across different types of schools. They sought general consensus on the goals and metrics, especially among school leaders. Thanks to this extensive engagement during all aspects of the process, implementation of the CSPF was smooth.

No one set of performance measures can address the specific needs or interests of every group, however. Both Chicago and Louisiana leaders made it clear that despite an extensive engagement process, it would be impossible to accommodate the preferences of every stakeholder.

As the accompanying Chicago case study notes, it’s wise to check in early with the people who have the ability to kill the effort to create a CSPF, know who is steadfastly in your corner, focus especially on the political system and its players, and know your base. It is impossible to engage too much given how much these schools have on the line.

2. Be realistic about the outcomes and metrics of the CSPF. The process of developing a CSPF can easily take on a life of its own, resulting in a tool that attempts to be everything to everyone or that tries to provide data for every possible question. For that reason, clarity on the goals of the CSPF is critical to success. The risk of extensive engagement is that the tool becomes watered down, rendering it useless. It takes strong leadership to keep the process focused for a strong end result. Some cities did this by getting community input on a strawman tool, others took community feedback and developed a tool that would meet basic community concerns. Both resulted in solid tools because the purpose was clear.

Apples to Apples: Common School Performance Frameworks as a Tool for Choice and Accountability 5

Just as important, the metrics used in the CSPF must be realistic. It must be feasible to collect the data, both technically and politically. Los Angeles and Louisiana leaders found that simple metrics are easier to agree on and require little to no policy changes. Leaders in both places saw the CSPF as an opportunity to make a coherent story out of the extensive data that are already collected, rather than finding ways to collect new data. At any cost, be sure that school staff believe in the core assessment data. That will make or break efforts to implement a CSPF.

3. Be purposeful and focused. Throughout the process, it is important for leaders of the process to communicate clearly how the framework will be used and what outcomes are expected. This helps reduce arguments and uncertainty. Transparency and effective communication about how the community’s values and priorities are being conveyed through the CSPF will support use and gain broader buy-in.

4. Be thoughtful and gradual on implementation. Give schools time to adjust to the new system and develop respect for it. A CSPF is useless if it is not used for decision making, but decisions will not be politically sustainable if the framework is not seen as valid, reliable, and sensible. Some cities gave schools their data ahead of time so schools could see how they would rank or compare in the new system. Others allowed charter schools to opt in rather than being forced to take part in the new system. Still others offered a year of “safe harbor” before the new rankings took effect. Across all of these examples, time spent letting schools work with and understand the data was time extremely well spent. The new frameworks in most of these cities were generally viewed as better than the old systems. And because no rating is ever perfect, most cities solicited early feedback and committed to making iterations and improvements over time.

Conclusion

Through CRPE’s research, we have found that CSPFs can be an essential tool for informed decision making, accountability, and parent information, but only if developed with diverse and strong buy-in and with a broad understanding of and commitment to how it will be used. The six cities referenced here (and many beyond those included here) shared a need to provide parents with reliable and consistent measures of school quality across all public schools. They also share difficult political tensions and the pernicious problem of how to sustain such efforts amidst leadership turnover and shifting priorities. So this work is not for the faint of heart and it is not a quick fix for either parent information gaps or holding

schools to high standards. In places that failed to fully implement the CSPF, the hurdles included changes in district leadership, stakeholder opposition, and poorly conducted community engagement. In those places that have continued to use a CSPF, there is an openness to iteration and a school board that has hired leaders who pledge to prioritize and preserve it.

As cities and states re-think their school accountability measures under ESSA, the observations and suggestions represented here provide direction to mitigate political opposition and build a tool that facilitates choice and is representative of the community’s values.

Apples to Apples: Common School Performance Frameworks as a Tool for Choice and Accountability 6

Table 1: Cities or Regions That Are Using, Have Attempted, or Are Developing a CSPF

Name of Framework Stage of DevelopmentFeatures and Coverage of Framework Purpose Contact Information

CHICAGO

School Quality Rating Policy (SQRP)

Complete; in use by Office of Accountability since the 2013-2014 school year.

Developed by Chicago Public Schools. Provides summative ratings for all CPS-run schools and all but one charter school.

Used for 5 purposes: parent information, identification of high-achieving schools, as a goal-setting framework, targeting assistance, guide Board decision making.

Office of Accountability [email protected]

DENVER

School Performance Framework (SPF)

Complete; in use since 2008.

Developed by Denver Public Schools. Over 80 metrics used. Rates all DPS-run schools and all charter schools.

Used for 5 purposes: diagnostic tool, performance transparency, district decision making, basis for accreditation, compensation decisions.

SPF, in the Academic and Innovation Office [email protected]

LOS ANGELES

School Performance Framework (SPF)

Complete; in use since Fall 2012.

Developed by Los Angeles Unified School District (LAUSD) to apply to all LAUSD-run schools, and independent and affiliated charter schools.

Developed with the intention of making previously gathered data useful for schools.

Used as a tool for school-level decision making, parent information, differentiated oversight of schools, charter renewal.

Division of Intensive Support and Intervention SchoolPerformanceFramework @lausd.net.

LOUISIANA STATE DEPARTMENT OF EDUCATION

Charter School Performance Compact (CSPC)

Complete; in use since the 2013-2014 school year.

The CSPC applies to all charter schools authorized by the Louisiana Board of Elementary and Secondary Education, including most schools in the Recovery School District.

Used as a tool for differentiated oversight of high- and low-performing schools, parent information, growth and renewal decisions.

Louisiana State Dept. of Education (877) 453-2721

MEMPHIS

Common information system: Memphis School Guide

In development; in initial attempts, leaders were not able to create a fully aligned CSPF. Memphis is still working towards a full CSPF.

A third-party group independently developed the Memphis School Guide parent information system, which is complete and in use.

Project was a collaborative effort between Shelby County Public Schools and the Achievement School District (ASD). Would have applied to all public schools in the city: traditional, charter, and ASD schools.

Memphis School Guide was developed by a parent group and encompasses all Memphis public schools, both district and charter.

CSPF was intended to be a consistent way to measure school quality.

The Memphis School Guide is a parent information system, to support school choice.

Memphis School Guide hello@ memphisschoolguide.org

SACRAMENTO

Guide to Success Incomplete; project was abandoned after leadership transition.

N/A Intended to be used as an accountability tool.

N/A

Apples to Apples: Common School Performance Frameworks as a Tool for Choice and Accountability 7

Name of Framework Stage of DevelopmentFeatures and Coverage of Framework Purpose Contact Information

BALTIMORE*

Two systems:

One for charter schools (School Effectiveness Review), one for district schools

Both frameworks are complete, and recently adopted.

Baltimore City Public Schools houses the frameworks in the Office of Achievement and Accountability. The frameworks used for district and charter schools are closely aligned.

School Effectiveness Review is an aspect of all Baltimore charter schools’ renewal process, and is a tool for differentiated oversight for all schools.

Achievement and Accountability Office (443) 984-2000

CLEVELAND*

School Quality Guide Development of common school profiles are complete, which include common academic metrics. First reports were in 2013.

The Cleveland Transformation Alliance, a public-private partnership, developed the School Quality Guide. The profiles are created for all Cleveland district schools and most charter schools.

The School Quality Guide is primarily a tool for parent and community information.

Cleveland Transformation Alliance (216) 592-2425

WASHINGTON, D.C.*

Common information through the D.C. Equity Reports and the Learn D.C. website

No common school ratings across district and charter schools, but the Equity Reports are complete and have been in use since the 2012–2013 school year.

Developed collaboratively between the Office of the State Superintendent of Education, the DC Public Charter School Board, and the Deputy Mayor for Education. Reports are created for all D.C. district and charter schools.

The Equity Reports are used primarily for parent and community information.

Office of the State Superintendent of Education [email protected]

NASHVILLE*

Two systems:

district (Academic Performance Framework) and charter

Both frameworks are complete. The Academic Performance Framework (APF) was adopted in 2013.

The APF was developed by Metro Nashville Public Schools (MNPS) and is applied to all MNPS-run schools. MNPS has an aligned framework used to evaluate charter schools.

The APF is used as a tool for differentiated oversight of schools and support to schools as well as parent and community information.

Metro Nashville Public Schools [email protected]

TULSA*

Tulsa Value-Added Reports

No common framework, but the district has worked with district-authorized charter schools to develop a framework for charter schools. Value-Added Reports are used for both district and charter schools.

Value-added school reports, developed by Tulsa Public Schools (TPS), are created for all district schools and district-authorized charter schools. There is interest in a CSPF, but it is not yet in development.

Value-added reports are used for informing district decision making, identifying best practices and low-performance.

Tulsa Public Schools valueadded@ tulsaschools.org

*These cities were not part of CRPE’s in-depth CSPF research.

From: Seth Litt <[email protected]> Date: September 30, 2019 at 10:10:08 PM PDT To: [email protected], "Crain, Jefferson" <[email protected]>, Kelly <[email protected]>, [email protected] Cc: "Vandenbos, Megan" <[email protected]>, [email protected], [email protected] Subject: Material Submission For Consideration at 10/8 Curriculum and Instruction Committee Hearing

Dr. McKenna, I am respectfully submitting the attached materials to you, as chair of the LAUSD Board Curriculum and Instruction Committee. I request that these documents be included in the materials distributed to all committee members for the 10/8 meeting and that these materials be taken into considered as part of the committee's discussion and recommendation vote on Board Vice President Goldberg's resolution, " Ensuring Support for Students at All Los Angeles Unified School District Schools" The attached documents are: 1) "Parents Deserve Clear Information About Student Growth in Schools" a new resource from National PTA and Data Quality Campaign. 2) "Growth Data: It Matters, and It’s Complicated" - a report from Data Quality Campaign that provides an overview of the ways that 48 states (excluding California and Kansas) are using student growth data. 3) A policy brief written by Morgan Polikoff, Ph.D., associate professor of education policy at the USC Rossier School of Education, co-director of the Center on Education Policy, Equity and Governance. This brief addressing myths about student growth measurements and the California dashboard. 4) A letter from 11 education organizations on the benefit and the need for release of student growth measurement. Thank you for your consideration and for your service to Los Angeles children. Seth

Document 1

Parents Deserve Clear Information About Student Growth in Schools • 1 •

INFORMATION

Student academic growth is an important data point that describes change in learning over time using annual test scores. In most states growth data is reported on school report cards—the online reports that show how your child’s school is supporting students.

When viewed side by side with data that is based on an annual test score, school-level growth data offers parents a more complete picture of school quality and improvement, and individual-level growth data provides them with an understanding of their child’s progress. Measuring growth acknowledges that a student’s starting point is as important as their ending point. But parents need to know that there are different ways to measure growth and that each method provides unique insights into school quality.

Do you know what student growth data is telling you about the quality of teaching and learning at your child’s school? This brief provides an introduction to the different ways states are using annual test scores to measure student academic success over time and steps you can take to understand growth at your child’s school.

About Student Growth in Schools

Parents Deserve CLEAR

• SEPTEMBER 2019 •

Parents Deserve Clear Information About Student Growth in Schools • 2 •

Public schools have a responsibility to deliver a quality education to all students. State leaders set goals for a quality education, and every year they check whether schools are meeting these expectations by looking at a variety of data about the schools. The Every Student Succeeds Act (ESSA)—the federal law that governs K–12 education—requires states to administer annual state assessments in reading and math in grades 3–8 and once in high school. The state must also administer a science assessment once in grades 3–5, once in grades 6–9, and once in grades 10–12.

X State leaders, superintendents, and school principals use these measures to hold schools responsible for helping students learn. Holding schools responsible is called school accountability, and it is required by ESSA.

X Accountability shows state and district leaders how all schools are performing and which schools need more support to help students learn.

X States must publish a public report card with information about how schools are serving students so that everyone, including parents and families, can understand school quality.

Two measures of academic success that state leaders look at to see whether schools are helping students learn are “proficiency” and “growth.” Both of these measures are based on annual reading and math test scores.

Proficiency rates help leaders understand how many students met a specific target in a given year, while growth data helps leaders understand student change in performance over time. These measures are most helpful to look at side by side.

How Do States UseANNUAL

TEST SCORES to Measure School Quality?

PROFICIENCYProficiency rates provide information about whether students are mastering academic skills. For example, do students in this school understand fractions and other math concepts the way third graders should?

• A proficiency rate tells you what percentage of students in a school have achieved “proficiency.” Achieving proficiency means students have met or exceeded a target score on the annual reading or math test that year. The proficiency target score is decided by your state.

• Every state is required to measure proficiency rates in reading and math for accountability under ESSA. You can find this proficiency data on your state report card.

STUDENT GROWTHStudent growth data describes change in learning over time. For example, did the students in this grade learn more, less, or the same about math this year?

• All growth data is based on two or more annual test scores (for example, Joey’s score on the annual test this year and his score last year).

• While not required, nearly every state uses student growth for accountability—but each state has its own way of calculating this data.

• It is important to remember that student growth data provides different insights into school quality depending on how it is calculated. (See “Quick Guide to Student Growth Data” on p. 4 for more on those different approaches.)

Parents Deserve Clear Information About Student Growth in Schools • 3 •

State TestResults

MARIA’S DASHBOARD

States have selected different ways to measure growth to show a more complete picture of student achievement. Some are using more than one method, which is helpful for understanding different aspects of students’ academic progress. Every state has its own way of calculating this data, which is why it is important to know your state’s unique approach to measuring student growth.

Why use school-level growth measures?Growth measures give state and local leaders a more complete picture of school quality and help shine a light on schools that are typically overlooked by proficiency rates alone. For example, imagine a superintendent sees that 20 schools in her district have low proficiency rates in English language arts, which means that most students at these schools are not reading and writing on grade level. This data tells her these schools are in need of additional support. But it doesn’t give her much information to act on.

She can also see the growth data for each school in her district. She can compare the data and see that one of those schools has high growth compared to the others. This information tells her that this school is doing a good job helping students improve, even though they are still performing below grade level.

She and her team might reach out to the principal to learn more about what teaching strategies the school is using. She can share those best practices with other principals. Without this growth measure, a state might overlook schools that are using promising practices because they have low proficiency rates. Growth measures provide a richer picture and show there is more to a school’s story.

Which growth measure is your state using?

To learn more, check out DQC’s resource:

Growth Data: It Matters, and It’s Complicated.

SCHOOL / DISTRICT DATA

Parents Deserve Clear Information About Student Growth in Schools • 4 •

Growth to proficiency measures look at how much students have or have not improved on

annual state tests in the past to predict when students will reach a set score that counts as proficient. These measures help identify schools where more students will reach or maintain proficiency.

Are students on track to reach proficiency within the next few years?

Gain score measures track student improvement on annual state tests year to year.

These measures demonstrate how many students in a school achieved the state target for improving their score.

Did students in this school improve their reading scores, even though they already achieved proficiency last year?

Value table measures create additional performance levels above and below

proficiency, such as “advanced” and “below basic.” These measures demonstrate whether students have improved, even if they haven’t reached the proficiency target.

Did students in this school learn enough to move their test scores from below basic to basic, even if they didn’t reach proficiency?

SGP measures demonstrate how students in a school performed compared to a group of other students in the same grade across the state who performed similarly in the past. These measures show differences in how schools are serving students at the same proficiency level.

Did students in this school do better than their peers even if they didn’t improve much?

Below are the five most common ways states are calculating growth data for accountability, including the technical term that may be used to describe the measure on a school report card.

Quick Guide toSTUDENT GROWTH DATA

There are different ways to measure growth, which give different insights into student success.

Of the five growth measures, value-added measures use the most sophisticated methods

to calculate growth. Value-added measures use information about a student’s past performance to make a personalized prediction for how they will perform on the annual test. Then, these measures compare the student’s actual score to their predicted score. The difference between these scores demonstrates how educators in a school affected student learning. While value-added measures can be more difficult to understand compared to other measures, they are the only measures that demonstrate how educators affect student performance.

Did teachers help students in this school do better than we expected them to perform, even if they didn’t get to a grade-level target?

VALUE ADDED VALUE TABLE

GAIN SCORE’14 ‘15 ’16 ’17

MATH

READING

GROWTH TO PROFICIENCY

STUDENT GROWTH PERCENTILE (SGP)

Each approach to measuring student growth gives a different insight into student achievement. What is student growth data telling you about your child’s school?

Parents Deserve Clear Information About Student Growth in Schools • 5 •

With growth data, you can gain a more complete picture of students’ academic success in your child’s school. But you shouldn’t have to be

a detective to find and understand this data.

Ask yourself the following questions:

� Do I know where to find the public report card for my child’s school?

� Can I easily find data about student growth or progress on the report card?

� Do I know what this data is telling me?

� What questions do I have about this data?

Take the following steps to gain a clearer understanding:

Ask your child’s principal or teacher questions such as the following:

� How is growth measured at this school? Does it offer information about student progress toward a learning goal? Does it measure how adults are helping students?

� What does this growth data reveal about student achievement and progress?

� How does this growth compare to other schools in the district?

� How do you and the school use this information to improve student success?

� Where can I access information about my child’s progress?

� How do you protect my child’s privacy?

Call on your state leaders to ensure that

� school-level growth data is easy to find and understand; and

� parents and educators have access to individual students’ growth data.

You Deserve CLEAR INFORMATIONAbout Student Growth on School Report Cards

SCHOOL PROFILES

DEMOGRAPHICS

ASSESSMENTPROGRAMS

FACILITIES

FUNDING

STRENGTHS: MATH

INTERESTS: PROGRAMMING

GOALS: COLLEGE

When you have the information you need about student growth at your child’s school, you can be a stronger advocate for your child.

Document 2

Growth Data: It Matters, and It’s Complicated

January 2019

Measures of student growth offer a richer understanding of student performance than a moment-in-time test score alone, but measures of student growth are not created equal. In their accountability plans for implementing the Every Student Succeeds Act (ESSA), some states are using a sophisticated analysis of multiple data points that evaluate the impact of schools on student learning, while others are using simpler measures of change in student assessment results year to year. These different approaches to measurement answer different questions and tell different stories about what is happening in schools and classrooms.

Leaders in all but two states have committed to measuring student growth as an indicator of academic achievement in accountability systems for elementary and middle schools, and 20 states will do so for high schools. States are responding to public demand for more information about school quality and student success that goes beyond a one-time test score. This development is made possible by years of state and federal investments in using longitudinal student data to draw richer insights about teaching and learning.

This data will be published on school report cards, which means that families, communities, and policymakers will have more information about student progress than ever before. To make sense of what the data on school report cards tells us about student success and school quality, and use it to inform decisions that improve student outcomes, everyone—from parents to policymakers—needs to be able to understand what their state is measuring and what it means about students’ academic progress. To offer some clarity, this brief provides an introduction to the different ways states will measure student growth in their ESSA accountability systems to evaluate school quality and support improvements.

Who should use this brief: Policymakers, such as legislators and board members, and state advocates who need to be informed about the ways that states are measuring student achievement and school quality can use this brief as a first step in understanding how student progress is captured in state accountability systems. Policymakers and advocates can also read about actions they can take once they have learned more about student growth.

What Is a Growth Measure? Most consumers of education data are familiar with status measures, often referred to as “proficiency.” Status tells us a student’s performance at one moment in time, usually based on a standardized test score. It tells us: Is Grace reading on grade level right now?

Growth measures use an individual student’s assessment data over time to evaluate some aspect of that student’s academic progress.

There are different ways to design growth measures. Some growth measures use advanced statistical methods; some use a more simple calculation. Each measure follows a set of processes or rules based on decisions made by people. Because they use different data and different methods, growth measures answer different questions, such as:

• How much has Grace learned since last year?

• How much has Grace learned compared to her similarly performing peers?

• How much has Grace’s school contributed to her learning?

For the purposes of this paper, all measures that use individual student assessment data over time to evaluate some aspect of student progress are “growth” measures, mirroring the language states are using. But it is important to note that, though the terminology is the same, “growth” means something different in each state, and the measures answer different questions.

Executive Summary

2 Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

Growth data offers a richer picture of student performance than a one-time test score alone.

Every student should make learning progress each year. By capturing the change in students’ test scores over time, growth data helps teachers and parents understand students’ progress, including whether each student is improving or falling behind. Growth data is based on assessments, but instead of just one score, measures of student growth use different methods to analyze individual students’ test scores over time. (See sidebar, “What Is a Growth Measure?”)

When considered side by side with other information, including proficiency measures, growth data provides a more comprehensive picture of student learning and a better indication of students’ progress than a one-time test score alone. Teachers and parents can act on this valuable information to support students on their path to college and career—it is especially important for supporting students who may not be meeting learning benchmarks but have still improved. District leaders can use this information to identify schools that are outperforming their peers in improving student achievement and dig into best practices. Community members can understand whether schools are helping every student make learning progress each year and advocate for supports and resources. State policymakers can use this information to identify which districts may be in need of additional support.

The Every Student Succeeds Act (ESSA) prompted more states to shine a light on student growth.

ESSA ushered in a new era of state education accountability and opened the door for an unprecedented number of states to include measures of student progress in accountability systems. Critically, ESSA moves away from a reliance on a one-time test score, or proficiency measure, as the primary measure of student success and school quality and requires states to include an additional measure of student achievement. For elementary and middle schools, this measure will be a measure of “student growth” if determined appropriate by the state or “another reliable

1 California is measuring school-level change in performance over time, which compares the performance of different groups of students (for example, this year’s third graders compared to last year’s third graders). Kansas is measuring achievement gaps.

statewide academic measure that allows for meaningful differentiation in school performance.” While stopping short of requiring states to include student growth, ESSA encourages states to do it—though the law does not define student growth or how states should measure it.

Based on their approved ESSA plans, all but two states—California and Kansas—have seized this opportunity to include a measure of “student growth” in their accountability systems.1 These measures, though different, evaluate the learning progress of individual students. Forty-eight states and Washington, DC, will use a measure of student growth for elementary and middle schools, and 20 states have committed to include such a measure for high schools.

This new data about student progress will appear on school report cards alongside other information about student success and school quality. As a result, families, communities, and policymakers will have more information about student progress than ever before.

The growth measures states have chosen for their accountability systems are not created equal.

While including student growth among a set of robust indicators used to assess school quality is an important step, consumers of this data need to know that states

Introduction

Longitudinal Data Systems Make Growth Data Possible Every state has the ability to use the most robust measures of student growth, such as value added and student growth percentile, because states have invested in statewide longitudinal data systems. These systems, which link state education data over time to provide a complete academic history for each student, are essential to calculating and using student growth measures. States have done the hard work to build these systems, establish relationships, and develop policies that support their ability to use this data to answer critical policy questions and support continuous improvement. Using student growth data to understand school quality is one way states are maximizing investments in longitudinal data.

3Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

are arriving at these indicators differently. States have made decisions about how to define, calculate, and include measures of student learning over time in their accountability systems. Ultimately these decisions have an impact on public understanding of school quality and the way that resources and supports are deployed to schools.

States are using the same language—“growth”—but different methods to calculate it. While most states are including an indicator of student growth in their accountability systems, each state is measuring different things that lead to different conclusions about student learning over time.

State leaders select indicators by considering what questions they want to answer, state goals, capacity, cost, ease of implementation, and feedback from stakeholders.

In addition to choosing the types of indicators they want to use, states also decide how to calculate, summarize, and interpret them. These differences matter.

The differences also mean that, while they might share common building blocks, every state’s approach to measuring student growth for accountability is ultimately different. Even if states use a common assessment, such as the Smarter Balanced Assessment Consortium exams, the growth data that states report by school cannot be compared across states.

This is complicated. If this data is going to deliver on the promise of giving everyone with a stake in education greater insight into student performance, everyone needs to understand their state’s growth measure and how the data can be interpreted.

Measures of Growth in ESSA 101

Measures of student progress referred to as student growth provide a richer picture than one-time test scores alone, but different growth measures accomplish different things. Most states are using at least one of five common types of measures of student growth, though each state has applied its own customizations:

(See the table for a high-level comparison of common growth measures, and see the map for state-by-state findings.)

X Value-added measures use advanced statistics and multiple data points to evaluate the impact of schools on student achievement.

X Student growth percentile measures use advanced statistics and students’ past performance data to evaluate how students are performing compared to their academic peers across the state.

X Value-table measures place students in performance levels based on their test scores and note when students move between levels year to year. These performance levels are a range of scores determined by the state.

X Gain-score measures use a change in test score on a comparable assessment year to year to demonstrate how much a student has learned over a given time period. An additional layer of analysis typically is applied to the test scores to make this measure possible; they are translated into what is called a “scale

score,” which allows for comparison (for example, if students take different versions of a test).

X Growth-to-standard measures evaluate the distance between a student’s current performance and a grade-level standard and, based on that student’s rate of progress, estimates how soon the student will meet that standard.

Value-added and student growth percentile measures offer insight into whether adults in a school are helping students learn, regardless of student proficiency level. These measures use advanced statistical methods to analyze students’ progress compared to what is normal or expected, providing for conclusions about a school’s collective contributions to student achievement. This calculation is not simple subtraction; the measures use multistep calculations that build on the capacity of state longitudinal data systems to support analysis of longitudinal student data and on years of work in the field to develop measures of school effectiveness using student performance data.

Value-table, gain-score, and growth-to-standard measures, on the other hand, track students’ mastery of state learning goals over time. Generally, these measures track the change in an individual student’s test score year to year and compare this change to state-determined expectations or other criteria to draw conclusions for accountability. (See table for a comparison.)

4 Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

Recognizing the differences between the types of measures and the kinds of insight they offer, 10 states are using multiple growth measures within their accountability systems, giving the states an even more robust picture of student progress.

As 48 states and Washington, DC, have included a measure of individual student growth alongside proficiency to

provide a richer picture of student achievement, it is important for consumers to know that the data is not created equal. Because these measures tell distinctly different stories about what is happening with student learning, states must communicate clearly which measures they are using. Each approach comes with tradeoffs for accountability, and states must be thoughtful to address each measure’s limitations.

Measures of Growth Used in State Accountability Systems

Type Value Added Student Growth Percentile Value Table Gain Score Growth to Standard

These measures use individual student performance data to demonstrate . . .

The impact of adults in a school on student achievement.

How schools served students with the same academic starting point.

Student progress. Student progress. A student’s distance from grade-level learning goals.

They are calculated by . . .

Using advanced statistics to analyze data about Joey’s past performance, and sometimes other characteristics that would affect his score, such as income or English language learner status, to predict how Joey will perform on the assessment.

Joey’s actual growth score is compared to his expected growth score, and the result is attributed his school.

Using data about Joey’s past performance, to group Joey with students across the state who got the same or similar score on the same test in the same grade.

Joey is then assigned a percentile or rank—between 1 and 99—based on how his current year performance compares to that of his academic peers.

Using a series of performance levels developed by the state that are based on a range of scores (e.g., 1–12 points, 13–24 points, etc.).

Joey’s test score this year is placed in a performance level and compared to where his score fell last year.

These measures note whether Joey moved between levels year to year.

Looking at the change in Joey’s test score on a comparable assessment from last year to this year. An additional layer of analysis is applied to the test scores to make this measure possible; they are translated into what is called a “scale score,” which allows for comparison (for example, if students take different versions of a test).

Comparing Joey’s performance this year to a long-term learning goal.

Assuming Joey will improve at the same rate every year, this type of measure estimates whether Joey is on track to achieve that goal within a given timeframe.

The resulting data tells us . . .

Joey’s school helped him improve more than other schools helped similar students.

Joey is in the 70th percentile; compared to a group of academically similar peers, he did better than 70 percent of them.

Joey moved from below basic to basic based on the state’s cut scores.

Joey scored 50 points higher than last year.

As a 4th grader, Joey is 100 points away from proficiency and is on track to be proficient by the end of the next two years.

5Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

Type Value Added Student Growth Percentile Value Table Gain Score Growth to Standard

Certain features of this type of measure have implications for accountability. These measures . . .

Can account for additional student background characteristics beyond academic performance.

Use advanced statistics and provide information that is different and complementary to a proficiency measure.

Can be challenging to communicate. Their sophistication can be challenging to explain to a broad audience.

Do not account for standards. These measures do not reveal any information about how students are performing relative to grade-level learning goals.

Use advanced statistics and provide information that is different and complementary to a proficiency measure.

Use only test scores. These measures do not account for other factors that contribute to student progress beyond a test score.

Do not account for standards. These measures do not reveal any information about how students are performing relative to grade-level learning goals.

Are provided in terms that are more familiar to many stakeholders. These measures use simple language that is commonly used in education, such as proficiency, which makes interpreting the change in a student’s mastery level easier.

Like proficiency measures, depend on state-determined criteria. Student performance depends on how high or low states have set accountability goals and expectations.

Do not account for school effectiveness independent of student proficiency status. These measures are unable to provide insight into how the adults in a school contributed to student performance.

Like proficiency measures, depend on state-determined criteria. The resulting growth data depends on how student test scores are translated into scale scores, which is part of the analysis that allows for comparison.

Do not account for school effectiveness independent of student proficiency status. These measures are unable to provide insight into how the adults in a school contributed to student performance.

Offer a picture of whether students are on track to meet academic benchmarks. These measures are helpful for offering more information about students who are already performing below a proficiency benchmark.

Like proficiency measures, depend on state cut scores. Student performance depends on how high or low states have set benchmarks.

Are based on an estimate. Generally, these measures are based on an assumption that a student will continue to make the same amount of improvement on his or her test scores year after year, regardless of his or her starting point or other characteristics.

6 Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

Which growth measures are states using?

Based on states’ approved ESSA plans, the map below shows the measures states have indicated they will use to evaluate student growth in elementary and middle schools.

The Data Quality Campaign (DQC) reviewed every state’s ESSA plan to determine whether the state planned to use a measure of student growth for school accountability. Based on our analysis, and consultation with experts and existing resources, we determined which type of growth measure each state is using. States’ choices about which growth measures to use in accountability may evolve or change in the course of implementation.

23 states are using student growth percentile. This measure is the most common.

12 states are using a value table.

10 states are using a growth-to-standard measure.

9 states are using a value-added measure.2

3 states are using a gain-score measure.

3 states are using a less common growth measure. Based on our review, these states are using a measure of individual student progress that, as described in the plan, cannot be classified as one of the more common measures above.

2 This map has been updated to reflect Louisiana’s multiple growth measures as value added and growth to standard.3 Washington, DC, Massachusetts, and Maryland are considering this approach; their plans state that they will consider including a growth-to-

standard measure in the future in addition to the student growth percentile measure they are already using.

10 states are using multiple measures. These states will use more than one measure to evaluate student growth, combining the measures in various ways. Five states are pairing a growth-to-standard measure with a student growth percentile measure, which will give them insight into both how students are performing compared to their academic peers and how they are progressing toward state standards.3

Value Added Student Growth Percentile

Value Table Gain Score Growth to Standard

Other NoneStudent Growth Percentile & Growth to Standard

Student Growth Percentile, Value Table,& Growth to Standard

Value Table& Value Added

Value Added& Student Growth Percentile

Value Added Student Growth Percentile

Value Table Gain Score Growth to Standard

Other NoneStudent Growth Percentile & Growth to Standard

Student Growth Percentile, Value Table,& Growth to Standard

Value Table& Value Added

Value Added& Student Growth Percentile

Value Table& Value Added

Value Added Student Growth Percentile

Value Table Gain Score Growth to Standard

Other NoneStudent Growth Percentile & Growth to Standard

Student Growth Percentile, Value Table,& Growth to Standard

Value Table& Value Added

Value Added& Student Growth Percentile

Value Table& Value Added

Value Added Student Growth Percentile

Value Table Gain Score Growth to Standard

Other NoneStudent Growth Percentile & Growth to Standard

Student Growth Percentile, Value Table,& Growth to Standard

Value Table& Value Added

Value Added& Student Growth Percentile

Value Table& Value Added

Value Added Student Growth Percentile

Value Table Gain Score Growth to Standard

Other NoneStudent Growth Percentile & Growth to Standard

Student Growth Percentile, Value Table,& Growth to Standard

Value Table& Value Added

Value Added& Student Growth Percentile

Value Table& Value Added

Value Added Student Growth Percentile

Value Table Gain Score Growth to Standard

Other NoneStudent Growth Percentile & Growth to Standard

Student Growth Percentile, Value Table,& Growth to Standard

Value Table& Value Added

Value Added& Student Growth Percentile

Value Table& Value Added

Value Added Student Growth Percentile

Value Table Gain Score Growth to Standard

Other NoneStudent Growth Percentile & Growth to Standard

Student Growth Percentile, Value Table,& Growth to Standard

Value Table& Value Added

Value Added& Student Growth Percentile

7Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

Start Conversations About Your State’s Growth Data

Forty-eight states and Washington, DC, have committed to holding elementary and middle schools accountable for improving individual student growth in addition to students meeting an academic proficiency benchmark. This data has the potential to shift thinking about school quality and student progress, equipping decisionmakers at all levels with the information they need to continuously improve in support of student success.

But the landscape of student growth measures in accountability systems is complicated. The only way this data will deliver on its promise to complement achievement data and broaden the picture of student success is if it is transparent, well communicated, and readily available to those closest to students so that they can use it to take action.

State policymakers should:

X Be transparent about what growth measure they are using and why. Everyone deserves to know how their public schools are doing, and states have a responsibility to present meaningful information to the public. State leaders should build trust by providing the context and rationale for the choices they have made.

X Ensure that those closest to students have secure access to the data they need about students’ academic growth. None of this work will matter if those closest to students do not have the data they need to move the needle in schools and classrooms. This information will help improve practice only if it is shared with those working closest to students, such as teachers and parents, and with students themselves. States must ensure that educators and decisionmakers at all levels have timely access to useful, secure student growth data they can act on to support all students on their path to success.

State policymakers and advocates should:

X Start conversations about the state’s growth data. Ask questions to unpack the decisions that shaped the state’s growth data. Help others, such as educators and community members, understand what decisions were made.

X Understand how growth data fits within the context of other accountability measures. Ensure that all of the data points in state accountability systems work together to create a complete picture of student success and school quality. Ask questions to understand why and how growth is weighted and used in conjunction with other measures of school quality.

X Observe how these measures impact decisionmaking and whether they evolve in the course of implementation. Once this data is reported, ask how the state is using these measures to identify challenges and drive improvements and how the state is adjusting these measures in the course of implementation.

Data tells a story, and the story is richer when it helps educators and leaders understand student learning over time, not just at one moment. Collecting and reporting student growth is just the beginning. Everyone using student growth data must understand the value and limitations of the information and, like with all data, consider it within the context of other available information. Doing this will ensure that, together, families, educators, and policymakers can put data to work for students.

Ask Questions About Student Growth DataStudent growth data used for accountability, and therefore to inform decisionmaking about school quality and drive continuous improvement, reflects a series of decisions made by people. To understand what the numbers are telling you and unpack the decisions that shaped your state’s growth data, ask questions such as:

• What are state leaders trying to learn about student progress?

• How was the growth measure selected? What were the key considerations?

• What are the expectations for student growth each year? How does the state determine these expectations?

• Does this measure tell us how well students are performing relative to the state’s academic standards?

• Is the state measuring growth the same way for both high- and low-performing students?

• Is the state measuring just students’ progress, or is it trying to isolate the impact of schools, taking into account the diversity of students they teach?

8 Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

Appendix

Methodology

DQC employed a systematic analysis to determine which states, as detailed in their ESSA accountability plans, will use a measure of student growth to calculate academic indicators and which specific measure(s) they will use. DQC developed a tracking spreadsheet with a list of common growth measures used for accountability, informed by subject matter experts and existing academic resources, to review each state’s approved ESSA plan.

In September 2018, DQC reviewed approved ESSA plans—which live on state websites—for all 50 states and Washington, DC, including Florida’s plan once it was approved on September 26, 2018. We read the sections describing the academic indicators used for accountability (in the ESSA plan template issued by the US Department of Education: Title I, Part A, Section 4, iv. a & b). DQC analyzed the language in these sections to determine whether states were using a measure of student growth in the calculation of an academic indicator and, if so, which type(s) of growth measure were being described. If the state named the measure, it was classified as that measure. If not, we made our best determination based on our definitions of each type. The culminating step involved categorizing states based on our determinations about their models as well as the total number of measures used in each state.

Acknowledgments

DQC would like to thank Chad Aldeman, senior associate partner, Bellwether Education Partners, and Steven Glazerman, senior fellow, Mathematica Policy Research, for their advisement on the contents of this brief.

Characterizations of common growth measures found in this resource were developed in consultation with experts in the field and existing literature on this topic. To learn more, explore the following resources about the application of student growth measures in accountability systems:

X The Council of Chief State School Officers’ Considerations for Including Growth in ESSA State Accountability Systems provides a resource to support state policymakers as they walk through steps for actualizing a growth model for their state accountability systems.

X The Education Trust’s Individual Student Growth factsheet outlines two main approaches to measuring growth and their implications for accountability.

X A Practitioners Guide to Growth Models provides comprehensive definitions of growth models.

9Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

Growth Measures State by State

Findings from DQC’s analysis of state-approved ESSA plans:

State

Growth measure state will use for accountability indicator, according to ESSA plan

The state includes a measure of individual student growth in accountability

For elementary and middle school For high school

Alabama Gain Score

Alaska Value Table  

ArizonaMultiple Measures: Student Growth Percentile, Growth to Standard  

Arkansas Value Added

California None    

Colorado Student Growth Percentile

Connecticut Growth to Standard  

Delaware Other

District of Columbia Student Growth Percentile

Florida Value Table

Georgia Student Growth Percentile

Hawaii Student Growth Percentile  

Idaho Growth to Standard  

Illinois Other  

IndianaMultiple Measures: Value Table, Student Growth Percentile, Growth to Standard  

Iowa Student Growth Percentile

Kansas None    

KentuckyMultiple Measures: Value Table, Growth to Standard  

LouisianaMultiple Measures: Value Added, Growth to Standard

Maine Value Table  

Maryland Student Growth Percentile  

Massachusetts Student Growth Percentile

MichiganMultiple Measures: Student Growth Percentile, Growth to Standard

Minnesota Value Table  

Mississippi Value Table

Missouri Value Added  

Montana Other  

Nebraska Value Table  

NevadaMultiple Measures: Student Growth Percentile, Growth to Standard  

New Hampshire Student Growth Percentile  

New Jersey Student Growth Percentile  

New MexicoMultiple Measures: Student Growth Percentile, Value Added

New York Student Growth Percentile  

North Carolina Value Added

10 Data Quality Campaign | GROWTH DATA: IT MATTERS, AND IT’S COMPLICATED

State

Growth measure state will use for accountability indicator, according to ESSA plan

The state includes a measure of individual student growth in accountability

For elementary and middle school For high school

North Dakota Gain Score

Ohio Value Added

Oklahoma Value Table  

Oregon Student Growth Percentile  

Pennsylvania Value Added

Rhode Island Student Growth Percentile

South Carolina Value Added  

South DakotaMultiple Measures: Student Growth Percentile, Growth to Standard  

Tennessee Multiple Measures: Value Added, Value Table

Texas Gain Score  

UtahMultiple Measures: Student Growth Percentile, Growth to Standard

Vermont Student Growth Percentile  

Virginia Value Table  

Washington Student Growth Percentile  

West Virginia Value Table  

Wisconsin Student Growth Percentile  

Wyoming Student Growth Percentile  

Total 49 20

The Data Quality Campaign is a nonprofit policy and advocacy organization leading the effort to bring every part of the education community together to empower educators, families, and policymakers with quality information to make decisions that ensure that students excel. For more information, go to www.dataqualitycampaign.org and follow us on Facebook and Twitter (@EdDataCampaign).

Document 3

September 2019

On Growth Models, Time for California to Show Some Improvement

Morgan Polikoff

alifornia is one of just two states (with Kansas) that does not use a student-level growth model to measure school performance. This brief lays out a number of common beliefs about growth models and provides evidence that these beliefs are inaccurate or unsupported. In so doing, the brief makes a positive case that the state should adopt such a model and replace the current “change” metric in the California School Dashboard. Two specific models—student-growth percentiles and residual-gain growth models—would be a dramatic improvement over what the state currently uses and would much more validly identify schools succeeding and in need of support.

C

On Growth Models, Time for California to Show Some Improvement2

Introduction

Educational accountability in California is in a new era. As the federal government has relaxed its requirements for consequential accountability, the state has taken a new approach to evaluating and supporting schools. In the past six years, California:

1. Retired the Academic Performance Index, which it had been using to rate schools for nearly 15 years.

2. Ended No Child Left Behind accountability, moving from a focus on rewards and sanctions to a model of continuous improvement.

3. Enacted the Local Control Funding Formula and Local Control Accountability Plans.

4. Rolled out the new California School Dashboard to report on school performance on multiple test-based and non-test indicators.

A look at the key indicators in the Dashboard illustrates California’s effort to consider not just the status of school performance, but also changes in school performance. Specifically, California has chosen to include in the Dashboard ratings a “cohort-change” model that compares this year’s average score in a school or district to last year’s average score.

For many reasons (which I describe below), it is important for the state to include a measure of growth in its accountability system. But is the cohort-change model the right choice? California is one of just two states (the other is Kansas) that does not calculate or report a student-level growth model (i.e., one based on comparing the growth in achievement of individual students from year to year). Should it? What are some of the key considerations in selecting a growth model, and how might California use what we already know from other states and from decades of research to make the best selection?

The purpose of this brief is to discuss the reasons why California should adopt a measure of student growth that aligns with what we know about the design of such measures and their use in accountability and continuous improvement systems. To do this, the brief presents a number of common misconceptions about growth models and dispels them using existing evidence. For the most part, when this brief talks about student growth models, it is referring to the Residual Gain Model1; when other models are discussed they are called out as such.

edpolicyinca.org

Policy Analysis for California Education

3

Growth Model Misconceptions

California is too different from other states to learn about growth models from research done elsewhere

Many people believe that technical research, such as growth-model research, must be context-specific to be relevant. The truth is that there is no reason to believe that the general findings from the technical literature on growth models are context-specific. Furthermore, growth-model research comes from a wide variety of contexts, many of which look like California.

There is a very large body of research on growth models2, and more research is being produced all the time. This research comes from states that look demographically similar to California (e.g., Texas3 and Florida4) and from states that look quite different (e.g., Tennessee5 and Missouri6). None of the recent reviews on the topic, nor any of the individual state-specific studies, provide any indication that the methodological recommendations they make are state- or context-specific.

Furthermore, though California often thinks of itself as being distinct from other states, there are many ways in which California is similar to other large, diverse states. Although California is the most populous state (enrolling approximately 6.2 million students), it is only 17 percent larger than Texas in student enrollment7. California is highly diverse, with only 24 percent white students and 59.8 percent African American and Latinx students, but Texas is just 28.5 percent white and 64.8 percent African American and Latinx. California has a large number of school districts (1,059), but Illinois has 970 and Texas has 1,241. California also has a large percentage of students eligible for free- and reduced-price lunch (58.9 percent), but Florida has 58.8 percent. In short, California looks very much like other large states along most any dimension. There is no reason to think California cannot learn about growth models, their designs, and their effects by drawing on data and lessons from other states.

California already has a growth model in the Dashboard

Many people might look at the California School Dashboard and see that it accounts for a school’s “change” from last year to this year in assigning a Dashboard rating; they might believe that this change score is equivalent to a growth model. The truth is that the change score is not a growth model—it is more commonly referred to as a “cohort change” model or an “improvement” model8. It is substantively different from (and inferior to) a growth model.

The difference between a growth model and this cohort change model is straightforward. A growth model tracks the performance of individual children from year to year, comparing the growth rates of children in different classrooms, schools, or districts.

On Growth Models, Time for California to Show Some Improvement4

In contrast, a cohort-change model like the one California currently uses compares this year’s students in a school or district to last year’s students. They are fundamentally different approaches to looking at the change in performance over time (they literally measure different things), so the current Dashboard measure is not a growth model.

Consider a middle school where the students come in at the 30th percentile in 6th grade, advance to the 50th percentile in 7th grade and the 70th percentile in 8th grade9. This school is doing phenomenal things for children—raising their achievement dramatically. A growth model, shown in Table 1, would reflect this impact and would show this school as a huge positive outlier. In contrast, if the arriving 6th graders stayed at the 30th percentile year after year (as they likely would, given the stable relationship of school-average poverty with achievement levels), the state’s cohort change model would show this school as middle-of-the-road.

Figure 1. How cohort change models fail to measure student growth

Even if the current change measure isn’t the same as a student-growth model, it tells us the same things

Many people might believe that the state’s cohort-change model and a true growth model might differ technically while still producing the same or very similar results. The truth is that cohort-change models and adjusted-gain growth models can (and often do) produce substantially different results because they measure fundamentally different things

There are many reasons why these models produce different results. One obvious reason is because cohorts of students can vary dramatically from year to year, so the groups of students being compared to one another in a cohort-change model are often

Highly Effective Middle School

Grade 6 Grade 7 Grade 8 Average

Year one

Year two

Year three

Year four

Year five

30

30

30

30

30

30

50

50

50

70

70

70

50.0

50.0

50.0

+20

+20

+20

+20

+20

+20

+20

Source: Albert Shanker Institute

edpolicyinca.org

Policy Analysis for California Education

5

very different from one another. While comprehensive California data are not available, it’s estimated that around 8-15 percent of children move out of a school or district in a given year (numbers are even greater in high-needs schools)10.

At a more basic level, it is clear to see that a school that causes student achievement to grow substantially from year to year (thus, performing well under a growth model), could have no or little change in the cohort-change model if the school enrolled similar kinds of students from year to year. The opposite is also true—a school in a gentrifying area could enroll more affluent children each year, causing it to look better in a cohort-change model even if it is not actually causing student achievement to grow.

To see how pronounced the differences in the approaches are, a recent analysis using data from the CORE districts compared school ratings based on the cohort-change model currently used to growth model results based on a student-level growth model11. The results, seen in Figure 2, showed that large proportions of schools identified as low-performing using a cohort-change model were actually high-growth schools (the top left corner). Similarly, many schools that were above average on the cohort-change model scored below average on the growth model (the bottom right corner). In this instance, it is not that both models are equally wrong, it’s that the cohort change model is giving more incorrect (i.e., less valid) signals about school performance.

Figure 2. Comparison of cohort-change model (X axis) and student growth percentile (y axis)

There’s no agreement among researchers on which growth models to use

Many people might think there are too many kinds of growth models with no agreement even among experts about which of the models is the best. The truth is that only a very small number of models are regarded as the “gold standard,” and the choice of one model over another is more about values and intended uses than it is about which model is the best.

Source: CORE Districts. 2018 analysis of residual gain student-level growth models.

around 8-15 percent of children move out of a school or district in a given year (numbers are even greater in high-needs schools)10.

At a more basic level, it is clear to see that a school that causes student achievement to grow substantially from year to year (thus, performing well under a growth model), could have no or little change in the cohort-change model if the school enrolled similar kinds of students from year to year. The opposite is also true—a school in a gentrifying area could enroll more affluent children each year, causing it to look better in a cohort-change model even if it is not actually causing student achievement to grow.

To see how pronounced the differences in the approaches are, a recent analysis using data from the CORE districts compared school ratings based on the cohort-change model currently used to growth model results based on a student-level growth model11. The results, seen in Figure 2, showed that large proportions of schools identified as low-performing using a cohort-change model were actually high-growth schools (the top left corner). Similarly, many schools that were above average on the cohort-change model scored below average on the growth model (the bottom right corner). In this instance, it is not that both models are equally wrong, it’s that the cohort change model is giving more incorrect (i.e., less valid) signals about school performance.

Figure 2. Comparison of cohort-change model (X axis) and student growth percentile (y axis)

Source: CORE Districts. 2018 analysis of residual gain student-level growth models.

There’s no agreement among researchers on which growth models to use

Many people might think there are too many kinds of growth models with no agreement even among experts about which of the models is the best. The truth is that only a

Commented [KB2]: DESIGNER - please redo this figure to make it easier to read.

Commented [HJH3R2]: Is this possible? What would you need to do so?

On Growth Models, Time for California to Show Some Improvement6

Under the Every Student Succeeds Act, 42 states are using just four kinds of models: a student-growth percentile model, a value table, a growth-to-standard approach, or a residual- gain/value-added model12. Of these, 31 states are using one or both of the closely-related13 student growth percentile and residual gain/value-added models. A variety of studies support the general conclusion that these kinds of models—regression-based models that determine how much better or worse children score on a test given their prior achievement (and possibly other variables)—are the most appropriate for making inferences about schools’ effects on student achievement14. Residual-gain models fare the strongest from a validity standpoint, while student- growth percentile models fare slightly worse on validity but may be more understandable by parents or educators. There are numerous available reviews of the evidence about the strengths and weaknesses of different models, but there is broad consensus among researchers who study growth models that these two approaches are the least biased and most accurate.

Controlling for student demographics in a growth model means we are setting different goals for different children

Some residual-gain models incorporate demographic information about students, including possibly their free- and reduced-price lunch status, EL status, disability status, etc. Many people might see models that include these predictors and interpret them to mean that the model is setting different targets for students based on these demographic variables. The truth is that models of this sort do compare children to other children with similar characteristics and prior achievement, but the decision about whether or not to control for these characteristics in the model is a discussion about values, not a technical consideration.

There are several key questions the state might consider in deciding whether to control for student demographics in their residual-gain models. One question is whether the state is interested in comparing schools that are similar in terms of the kinds of students they serve. Put another way, should schools be punished or rewarded based on who the children are who happen to enroll at the school, or should comparisons be based on schools’ effects on those students’ performance? Similarly, should schools be compared fairly with themselves over time? For example, schools in rapidly gentrifying urban areas might quickly appear to be more effective because their student body is becoming more affluent. Controlling for student demographics would ensure schools are not benefiting from or being punished for demographic changes that are out of their control.

If the state is indeed interested in comparing schools net of their students’ characteristics, then a second question is whether the state is more comfortable under-correcting or under-correcting for student characteristics. The answer to this question could inform the specific type of residual-gain model chosen15. These choices are mostly

edpolicyinca.org

Policy Analysis for California Education

7

conceptual and value-driven, however, the actual differences in the performance of these models are modest and depend on the particular covariates used and how they are included in the model16.

“VAM is a sham”: These models don’t provide school effectiveness data that could help us make valid judgments about school effectiveness

Many people have heard that residual-gain and other growth-based approaches to measuring effectiveness are biased and invalid. The truth is that the best and most recent research concludes that growth-based estimates of effectiveness using residual-gain models demonstrate real and educationally meaningful differences in effectiveness that persist for many years. Furthermore, the most sophisticated residual gain models can all but eliminate concerns about bias. Finally, the question of validity and bias is a question about the use of the results—if the results are to be used for continuous improvement and other low-stakes purposes, validity and bias concerns are dramatically reduced anyway.

Two kinds of recent studies provide evidence that the best residual gain models produce either unbiased estimates of impacts on student achievement or estimates with extremely small bias. While these studies are in the context of teacher value-added, not school value-added (there has been much less research on the latter), there is no reason to think that the general findings do not apply. In one kind of study, students are randomly assigned to teachers in order to estimate teachers’ true impact on student achievement and compare it to estimates calculated from longitudinal data obtained prior to the random assignment17. In another kind of study, researchers use large-scale longitudinal data to relate estimates of teacher effects to students’ long-term outcomes18. The conclusion from these studies is that “estimates of teacher value-added from standard models are not meaningfully biased by student-teacher sorting along observed or unobserved dimensions19.”

Another important dimension of the validity question is “compared to what?” Currently, schools are evaluated based on their performance levels and the aforementioned cohort-change score. A wide variety of evidence makes clear that performance levels are largely a measure of who enrolls in a school (poverty and other demographic characteristics) and have little to nothing to do with school effectiveness. Figure 320 demonstrates this, showing the very strong relationship between school percent free and reduced lunch and school-average achievement levels (Figure 421 shows that this relationship is almost nonexistent for student-growth percentiles). And, as discussed above, cohort change measures have highly questionable validity as measures of effectiveness. Thus, even if there are modest questions about bias and validity with respect to the use of residual-gain growth models to identify the effectiveness of schools, there is absolute certainty these models are better from a bias and validity perspective than what the state currently uses.

On Growth Models, Time for California to Show Some Improvement8

Figure 3. Student achievement levels and school average income in a large national sample

Figure 4. Student achievement levels and school average income in a large national sample

Figure 3. Student achievement levels and school average income in a large national sample

Source: Evaluating the relationships between poverty and school performance

Figure 4. Student achievement growth and school average income in a large national sample

Commented [KB4]: DESIGNER - please redo Figures 3 and 4 to make them easier to read.

Commented [HJH5R4]: Is this possible? What would you need to do so?

Source: Evaluating the relationships between poverty and school performance

Growth models are too technical for educators or parents to understand

Many people might believe that growth models—especially residual-gain models that require statistical modeling—are too technical for educators or parents to understand or make use of. The truth is that educators and parents value growth data, and we have learned a great deal about how to present these data to stakeholders in ways that they understand and can use.

The first important fact is that educational stakeholders value growth data. In fact, recent research examining how parents make judgments about school quality found that parents place more weight on student growth than they do on achievement levels or any other criterion when comparing schools to make judgments about school quality22. Teachers also often value these data and, when properly supported and trained in how to analyze them, can use them to improve teaching and learning23.

Not only do these stakeholder groups value growth data, but they can be supported to help understand and make use of these data. There are numerous practitioner- and parent-oriented materials available to help users make sense of these data, ranging from whole books24 to reports25 to short briefs26. It is true that growth data, especially those based on residual gain and other advanced statistical approaches, are complicated. But experience makes clear that everyone can be made to understand these data.

Fall Reading Achivement for Over 1500 Publaic Schools Across the US

Fall to Spring Reading Growth for Over 1500 Public Schools Across the US

School Percent Free and Reduced Lunch

School Percent Free and Reduced Lunch

Med

ian

Fal

l Stu

den

t A

ch

ivem

en

t

Nat

ion

al P

erc

en

tile

Med

ian

Nat

ion

al S

tud

en

t

Gro

wth

Perc

en

tile

Source: Evaluating the relationships between poverty and school performance

Source: Evaluating the relationships between poverty and school performance

edpolicyinca.org

Policy Analysis for California Education

9

Growth models are too technical for educators or parents to understand

Many people might believe that growth models—especially residual-gain models that require statistical modeling—are too technical for educators or parents to understand or make use of. The truth is that educators and parents value growth data, and we have learned a great deal about how to present these data to stakeholders in ways that they understand and can use.

The first important fact is that educational stakeholders value growth data. In fact, recent research examining how parents make judgments about school quality found that parents place more weight on student growth than they do on achievement levels or any other criterion when comparing schools to make judgments about school quality22. Teachers also often value these data and, when properly supported and trained in how to analyze them, can use them to improve teaching and learning23.

Not only do these stakeholder groups value growth data, but they can be supported to help understand and make use of these data. There are numerous practitioner- and parent-oriented materials available to help users make sense of these data, ranging from whole books24 to reports25 to short briefs26. It is true that growth data, especially those based on residual gain and other advanced statistical approaches, are complicated. But experience makes clear that everyone can be made to understand these

data.

Growth models don’t make sense in an accountability and continuous improvement system

Many people think that the data we already have in the Dashboard is sufficient for California’s continuous improvement efforts. The truth is that the existing data, especially the cohort-change data, are insufficient for the task of contributing to continuous improvement.

In order for continuous improvement to succeed, we must first have an accurate sense of how schools are doing, where areas of need are located, and what practices predict improvements in outcomes. Simply put, we cannot have an accurate sense of any of these things if we do not have accurate growth data, and the current Dashboard measures do not provide accurate data on schools’ effects on student learning.

Using a growth model puts teachers at risk of being fired

Many people think that because growth models have been part of high-stakes teacher evaluation systems in other states, their use in California’s school accountability would lead to teacher evaluation reforms here. The truth is that creating a growth model has no bearing at all on the policy decision of how growth-model data are to be used. Furthermore, there is absolutely no indication that high-stakes teacher evaluation is on the

On Growth Models, Time for California to Show Some Improvement10

policy agenda in Sacramento, especially given California was one of the very few states that successfully resisted Obama-era encouragement to establish these systems.

Recommendations

Based on the existing literature and an examination of California’s own goals for the Dashboard and the continuous improvement system, the state should adopt a student-level growth model as soon as possible. Forty-eight states have already done so; there is no reason for California to hang back with Kansas while other states use growth data to improve their schools.

As described above, there are just a few kinds of growth models that are used in most states. Of these, there are two possible models that are most suited to California: student-growth percentiles (SGPs) and residual-gain models. Residual-gain models can be further categorized as one-step or two-step models, as well as models that do and don’t control for demographics other than prior achievement. A full review of these models is available elsewhere27, but this brief concludes with a short discussion of the strengths and weaknesses of these approaches.

Student Growth Percentiles

Student-growth percentiles are the most widely used growth models in state accountability systems. Student-growth percentiles use students’ prior test score history to answer the question “How well is this student doing this year compared to students with similar prior test scores?” SGPs are typically expressed in percentiles, so a score of 80 means that a student is doing better than 80% of students with her similar prior test history. SGPs can be averaged and reported for schools or districts. A major advantage of this approach relative to residual-gain models seems to be that it is relatively easier for practitioners to understand insofar as the numbers have a clear meaning. Disadvantages seem to be that the model may be subject to a bit more bias than residual gain models and that it has not been studied as much, so its properties are generally less well known. This model also has some technical downsides relating to its assumptions, but these are fairly typical of all approaches to measuring growth.

Residual Gain Models

SGPs are actually special cases of residual-gain models, which are also widely used by states. These models use students’ prior achievement, sometimes with additional demographic or other covariates, to answer the question “How far above or below expectation is this student performing given her prior achievement (and perhaps also given her demographics and that of the school)?” Residual gain models are by far the most common models in research—they are often used, for instance, in experimental evaluations of the impact of a given treatment when student achievement is also

edpolicyinca.org

Policy Analysis for California Education

11

measured. The research consensus is that these models (especially certain kinds of residual gain models) exhibit the least bias of all available growth models. They have similar technical limitations to student-growth percentiles, and they also may be somewhat more difficult to explain because they do not produce results on the percentile scale (though they can in fact be reported on a similar percentile scale to SGPs).

A separate decision, if a residual gain model is chosen, is whether or not to adjust the model for student demographics. Most researchers who study value-added models, and are concerned most about bias and the incentives inherent in choosing a model, would prefer a model that does control for students’ individual and peer demographics in addition to prior test scores. For example, Castellano and Ho (2013) argue, “If it seems that more grades [of prior achievement data] allow for an improved definition of academic peers, then why not improve the definition further by including demographic variables?28” Koedel and colleagues (2015) similarly argue that “in policy applications it may be desirable to include demographic and socioeconomic controls in [residual-gain models], despite their limited impact on the whole, in order to guard against [schools] in the most disparate circumstances being systematically identified as over- or under-performing.29” However, Koedel and colleagues also argue that the practical significance of not controlling for demographic variables is likely small—the correlation between results from models that do and do not control for demographic variables is typically close to 1.

Conclusion

Many people think that California’s current Dashboard and continuous improvement system represent a dramatic change over what it replaced. The truth is that California could very easily, and at close to zero cost, choose a growth model that would represent a dramatic improvement over what currently exists. While the residual-gain model is the consensus choice of most experts, even the student-growth percentile would be a fine choice. Either way, there is more than enough information for leaders in the state, including the State Board of Education, to make a decision, and they should act now.

Author Biography

Morgan Polikoff, Ph.D., is an associate professor of education policy at the USC Rossier School of Education, the co-director of the Center on Education Policy, Equity and Governance; and the co-editor of Educational Evaluation and Policy Analysis. He studies the design, implementation, and effects of standards, assessment, and accountability policies.

On Growth Models, Time for California to Show Some Improvement12

Endnotes1 See Castellano, K. E., & Ho, A.D. (2013). A practitioner’s guide to growth models. Washington, DC: Council of ChiefState School

Officers. Retrieved from https://scholar.harvard.edu/files/andrewho/files/a_pracitioners_guide_to_growth_models.pdf

2 For a nice overview see Castellano, K. E., & Ho, A.D. (2013). Ibid. Also see Koedel, C., Mihaly, K., & Rockoff, J. E. (2015). Value-added modeling: A review. Economics of Education Review, 47, 180-195. doi.org/10.1016/j.econedurev.2015.01.006

3 Lincove, J. A., Osborne, C., Dillon, A., & Mills, N. (2014). The politics and statistics of value-added modeling for accountability of teacher preparation programs. Journal of Teacher Education, 65(1), 24-38. doi.org/10.1177/0022487113504108

4 Winters, M. A., Dixon, B. L., & Greene, J. P. (2012). Observed characteristics and teacher quality: Impacts of sample selection on a value added model. Economics of Education Review, 31(1), 19-32. doi.org/10.1016/j.econedurev.2011.07.014

5 Sanders, W. L., & Horn, S. P. (1998). Research findings from the Tennessee Value-Added Assessment System (TVAAS) database: Implications for educational evaluation and research. Journal of Personnel Evaluation in Education, 12(3), 247-256. doi.org/10.1023/A:1008067210518

6 Ehlert, M., Koedel, C., Parsons, E., & Podgursky, M. J. (2014). The sensitivity of value-added estimates to specification adjustments: Evidence from school- and teacher-level models in Missouri. Statistics and Public Policy, 1(1), 19-27. doi.org/10.1080/2330443X.2013.856152

7 All statistics in this paragraph come from the 2017 Digest of Education Statistics, available online at https://nces.ed.gov/programs/digest/.

8 For more information, see Castellano & Ho, 2013. Op. cit.9 Example and figure drawn from http://www.shankerinstitute.org/blog/when-growth-isnt-really-growth.10 Welsh, R. O. (2017). School hopscotch: A comprehensive review of K–12 student mobility in the United States. Review of

Educational Research, 87(3), 475-511. doi:10.3102/0034654316672068 11 CORE Districts. (2018). Analysis of residual gain student-level growth models. Sacramento, CA.12 Data Quality Campaign. (2019). Growth data: It matters, and it’s complicated. Washington, DC.13 For a clear description of the similarities and differences between these two models, see Castellano and Ho, 2013. Op. cit.14 Koedel, C., Mihaly, K., & Rockoff, J. E. (2015). Value-added modeling: A review. Economics of Education Review, 47, 180-195.

Op. cit.15 See Ehlert, M., Koedel, C., Parsons, E., & Podgursky, M. (2014). Choosing the right growth measure. Education Next, 14(2), 67-

72. Retrieved from https://www.educationnext.org/choosing-the-right-growth-measure/

16 Parsons, E., Koedel, C., & Tan, L. (2019). Accounting for student disadvantage in value-added models. Journal of Educational and Behavioral Statistics, 44(2), 144–179. doi:10.3102/1076998618803889

17 See Kane, T. J., McCaffrey, D. F., Miller, T., & Staiger, D. O. (2013). Have we identified effective teachers? Validating measures of effective teaching using random assignment. Seattle, WA: Bill and Melinda Gates Foundation. See also Kane, T. J. & Staiger, D. O. (2008). Estimating teacher impacts on student achievement: An experimental evaluation. NBER Working Paper No. 14607.

18 Chetty, R., Friedman, J. N., & Rockoff, J. E. (2014a). Measuring the impacts of teachers I: Evaluating bias in teacher value-added estimates. American Economic Review, 104(9), 2593-2632. Retrieved from http://www.jstor.org/stable/43495327. Chetty, R., Friedman, J. N., & Rockoff, J. E. (2014b). Measuring the impacts of teachers II: Teacher value-added and student outcomes in adulthood. American Economic Review, 104(9), 2633-79. doi:10.3386/w19424.

19 Koedel, C., Mihaly, K., & Rockoff, J. E. (2015). Value-added modeling: A review. Economics of Education Review, 47, 180-195. Op. cit.

20 Figure 2.1 from Hegedus, A. (2018). Evaluating the relationships between poverty and school performance. Portland, OR: NWEA.

21 Figure 2.2 from Hegedus, A. (2018). Evaluating the relationships between poverty and school performance. Portland, OR: NWEA.

22 Korn, S.A. (2018, March). Would you recognize a quality school if you saw one?: Exploring parents’ evaluation of schools using Mechanical Turk. Paper session presented at the 44th annual meeting of the Association for Education Finance and Policy Conference, Kansas City, MO. 

23 Education Trust. (2013). The value of value-added data. Washington, DC: Author. Tennessee SCORE. (2018). What teachers say about TVAAS. Nashville, TN: Author.

24 Harris, D. N. (2011). Value-added measures in education: What every educator needs to know. Cambridge, MA: Harvard Education Press.

25 Castellano, K. E., & Ho, A.D. (2013). A practitioner’s guide to growth models. Washington, DC: Council of Chief State School Officers.

26 Data Quality Campaign. (2019). Growth data: It matters, and it’s complicated. Washington, DC: Author.27 Castellano, K. E., & Ho, A.D. (2013). A practitioner’s guide to growth models. Washington, DC: Council of Chief State School

Officers. Op. cit.28 Castellano, K. E., & Ho, A.D. (2013). A practitioner’s guide to growth models. Washington, DC: Council of Chief State School

Officers. Op. cit.29 Koedel, C., Mihaly, K., & Rockoff, J. E. (2015). Value-added modeling: A review. Economics of Education Review, 47, 180-195.

Op. Cit.

Stanford Graduate School of Education

520 Galvez Mall, CERAS 401

Stanford, CA 94305-3001

Phone: (650) 724-2832

Fax: (650) 723-9931

edpolicyinca.org

Policy Analysis for California Education (PACE)

Policy Analysis for California Education (PACE) is an independent, non-partisan research center led by faculty directors at Stanford University, the University of Southern California, the University of California Davis, the University of California Los Angeles, and the University of California Berkeley. PACE seeks to define and sustain a long-term strategy for comprehensive policy reform and continuous improvement in performance at all levels of California’s education system, from early childhood to postsecondary education and training. PACE bridges the gap between research and policy, working with scholars from California’s leading universities and with state and local policymakers to increase the impact of academic research on educational policy in California.

Founded in 1983, PACE

• Publishes policy briefs, research reports, and working papers that address key policy issues in California’s education system.

• Convenes seminars and briefings that make current research accessible to policy audiences throughout California.

• Provides expert testimony on educational issues to legislative committees and other policy audiences.

• Works with local school districts and professional associations on projects aimed at supporting policy innovation, data use, and rigorous evaluation.

Recent PublicationsPolikoff, M. S. (2019). Gauging the revised California School Dashboard. Palo Alto: Policy Analysis for California Education.

Phillips, M., Reber, S., & Rothstein, J. (2018). Making California data more useful for educational improvement. Palo Alto: Getting Down to Facts II, Policy Analysis for California Education and Stanford University.

Hough, H., Byun, E., & Mulfinger, L. (2018). Using data for improvement: Learning from the CORE Data Collaborative. Palo Alto: Getting Down to Facts II, Policy Analysis for California Education and Stanford University.

Koppich, J. E., White, E., Kim, S., Lauck, M., Bookman, N., & Venezia, A. (2019). Developing a comprehensive data system to further continuous improvement in California. Palo Alto: Policy Analysis for California Education.

Polikoff, M., Korn, S., & McFall, R. (2018). In need of improvement? Assessing the California Dashboard after one year. Palo Alto: Getting Down to Facts II, Policy Analysis for California Education and Stanford University.

Document 4

9/24/2019

To Members of the Los Angeles Unified School District Board of Education and Los Angeles Unified School District Superintendent Austin Beutner,

Our organizations have been supportive of Los Angeles Unified School District’s initiative to create a school performance framework because our members know that access to clear information about the schools in their communities is a matter of equity.

Our intent in this letter is not to advocate for every element of the School Performance Framework, as it was described by the original resolution and designed by the LAUSD working group. Instead, we urge the school board and district leadership to release the measures of student academic growth before the opening of the eChoices window on October 1st, 2019. Our understanding is that this data is available and complete, making its release to educators, families and the general public solely dependent on the decision-making of district leadership.

Why Student Academic Growth Matters

California is one of only two states in the country (the other is Kansas) that does not measure and publicize student growth. The California School Dashboard measures only student academic proficiency. “Change” as measured on the California Dashboard does not measure the progress that students are making, it only measures the average achievement of one year’s group of students versus the previous year’s students.

This type of measurement is extremely unfair to schools that serve concentrations of high-needs students who enter the school already below standards. As an example, a middle school that enrolls many 6th graders who are far below standards in ELA and that helps its students make dramatic academic gains, so that most of its 8th graders are meeting standards, will not receive any recognition for this work since a new cohort of struggling 6th graders enters the school every year.

A measure of student academic growth tells us something much more important. It tells us the amount of academic progress that students make at each school. In a school system where almost 60% of students are not meeting standards in English Language Arts, nothing is as important as understanding which schools are doing the best job at helping students catch up. In fact, one can see the entire enterprise of LAUSD, a school system that almost exclusively serves students who start out behind their more privileged peers, as the work of helping children catch up and then excel.

The Impact of a Student Growth Measure on Schools

We believe that the release of student growth measurements will have significant benefit for Los Angeles Unified schools by keeping the focus where it should be: on accelerating student learning and closing achievement gaps.

LAUSD has some schools where less than 15% of students enter the school meeting academic standards. Most of these are neighborhood, zoned schools, that are not a part of any of LAUSD’s choice programs. For these schools, proficiency rates do not provide a fair or accurate evaluation of school quality.

!1

Amongst this set of schools, there are schools that are doing an excellent job of helping their students make academic progress. In our current reality these schools are not recognized for the work that they do. With the release of a student growth measure, these schools would be recognized for the progress that their students make.

Parents would choose to send their children to these schools. Far more than the governance structures of magnet or charter schools, families are looking for schools that will help their children make progress from where they currently are to their full potential. With increased enrollment, schools can maintain stable programs and staffing that would allow them to continue and accelerate their success. School stability and success, in turn, makes it easier to retain and to recruit staff, including teachers.

The Utility of a Student Growth Measure for Local Districts, Communities of Schools, School Board and Central District

Los Angeles schools unquestionably operate in a severely resource-constrained capacity, serving students whose needs are not matched by the financial resources that schools have at their disposal. The signers of this letter will continue to support LAUSD in efforts to raise student funding. At the same time, we believe that decisions on how to use resources, both monetary, as well as the knowledge and skills of district employees, are crucial to improving student outcomes.

A student growth measure is not the only place where those tasked with supporting schools should focus their work. However, a student growth measure, as designed for the SPF, would provide an objective starting place to understand all schools. Currently many school level plans, in an attempt to provide more accuracy and depth than SBAC proficiency rates, use measures of student progress that may not be the same as comparable schools.

The Utility of Student Growth Measure For Families, Especially Low-Income Families

Between charter schools and LAUSD choice programs, more than half of the students who attend public schools within the boundaries of LAUSD are enrolled through some type of parental choice. However, there are negative impacts when a high-choice system is married to a low-information environment.

Currently, families are overwhelmingly using SBAC proficiency rates, whether they are looking at Great Schools, the CA Dashboard, or LAUSD’s own school-finder tool. The school with the highest proficiency rate may not be the best school for an individual child, especially if that school involves hours of travel each day. In our work with families, especially those with middle or high school students, we frequently hear that parents want to find the schools that will help their children catch up if they have fallen behind. Without a student growth measure, families have no access to this most important question and cannot make the best choice for their children.

In our current system we also have a pattern where families with the most privilege can successfully navigate the system to secure spots in the “best schools” for their children. If the “best schools” are defined by proficiency rates, which are in turn largely a measure of how prepared each school’s incoming students are, and a proxy for income, then we have a system that perfectly sorts and segregates students. A measure of student academic growth is a powerful way to disrupt this pattern.

!2

We want to be clear that this is not a case for desegregating schools or a case for more choice. Those are separate and important conversations that deserve their own space. We are making a more practical argument, which is that right now, a majority of LA’s families are choosing their children’s schools based on limited information. It does not have to be that way.

A measurement of student growth would also allow families to more effectively partner with their current schools. Parents use lots of information to understand how well a school is educating their child. If parents only see proficiency rates as an indicator of academic quality, they are being given an inaccurate view of their schools. In some cases, families may assume that a high proficiency rate means that all is well. In other cases, they may think a school is not doing well, when student growth would show that the school community is making significant progress.

The Power of Student Growth to Build Public Support for Public Education

The last year has produced contradictory messages on how much the Los Angeles public supports public education, and the limits of that support. While public opinion was supportive of teachers during the UTLA strike, a few months later the public did not vote to fund public schools through the Measure EE parcel tax. If public schools are to fulfill their promise, the broader public needs to act as if public schools and public school students are their responsibility.

We have heard the argument that the SPF, including its student growth measure, are unnecessary because schools have all of the information that they need to make instructional decisions. However, educators are not the only audience for school data. The public at-large is, as well. It is members of the public who vote for school boards, to fund or not fund schools, and evaluate whether or not they will send their children to public schools.

There are many barriers to communicating both the real challenges and the real successes of Los Angeles schools. To galvanize public support, we must make clear the high level of need that LAUSD’s students bring to the classroom and the daunting challenge that LA’s educators face in meeting those needs. However, we should not make the task seem hopeless or impossible. It is not impossible. All over the district, there are amazing educators and extraordinary school communities, who despite significant challenges, are making meaningful gains with their students. Student growth data would allow educational leaders to tell a success story about many of the district’s schools and to establish a compelling vision for what would be possible with more resources to provide more students with these opportunities.

The Case for Releasing Growth Measures Before October 1st, 2019

We are aware that the full SPF, including the measure of academic growth has been completed by CORE and provided to LAUSD. While district staff can see every school’s growth measure, parents cannot. On October 1st, LAUSD’s E-choices portal will open and families all over the district will have a short period of time to make one of the most important decisions of their child’s life: where to send them to school. In this letter we have already laid out the consequences, especially for low-income families, of choosing schools without access to accurate information. Families in Los Angeles have had to do this for years.

!3

This can stop now. LAUSD has the information and must release it to families and include it on its school finder tool and the open data portal.

The second reason for releasing the student growth measures now is a matter of trust. Families advocated for the passage of the school performance framework. They attended and participated in the working group, even though the process took a year longer than what was called for in the resolution. They attended meetings with senior district staff to make sure that the process was on track. What does it say to them, when a few weeks from the release of information about their schools, and after having participated with good faith in the process, they will no longer get the information that they sought?

There may be a reasonable debate about the value of overall ratings, but research has clearly demonstrated the value of measuring student growth. Withholding this information further confirms for many families that LAUSD is not a place where they are heard and that, even when they use their time and their voices, they are still at the mercy of forces more powerful than them. This is a terrible and unnecessary message to send. LAUSD should not be taking the extraordinary step of preventing its families from having access to information which almost every other family in America has. We ask you to immediately release the student growth measure and to include it on the LAUSD school finder tool, in the open data portal, and in other avenues of parent engagement. We look forward to your response,

!4

Seth Litt Executive Director

Joan Sullivan Chief Executive Officer

Layla Avila Chief Executive Officer

Ray Lopez Chang Program Officer, Education

Policy and Advocacy

Desiree Martinez Associate Director of

Organizing

Katie Braude Executive Director

Vanessa Aramayo Executive Director

Sarah Lillis Executive Director

Hannah Gravette Regional Vice-President,

Los Angeles

Oscar E. Cruz President & CEO

Ana Ponce Executive Director