CDE releases first set of growth model data

The California Department of Education on Sept. 22 released the initial set of historical student growth data on student achievement for English language arts and math on its Growth Model webpage.

The new model aggregates and reports data by local educational agency and student group, including racial and ethnic categories and other student characteristics. A score of 100 means that, on average, students are meeting their expected growth. A score below 100 means that, on average, students scored lower than their expected growth and vice versa.

Of the 16 different student groups that CDE released data on, on average, half of those groups scored lower than their expected growth. African American, American Indian, Pacific Islander, homeless and foster youth students were among the lowest average growth scores.

Complicating these averages is the number of LEAs with no reported data for specific student groups. This is partly because aggregate growth scores are not calculated for student groups with less than 11 individual student growth scores in the reporting year. If an LEA has fewer than 11 individual student growth scores, they do not receive a reported growth score within the data.

It is also important to note that averages of individual student groups represent aggregated scores across the state, thus masking some of the disparities across LEAs. Some of the most significant disparities in growth scores within student groups occurred with English learners and students with disabilities. Even though this data is informative, CDE stresses that it is not actionable because it is based on historical assessments. However, individual district data could be helpful for board members in identifying student groups that might need more robust educational supports.

The model
The “residual gain” model calculates differences between students’ predicted test scores and actual test scores, using previous English language arts and math scores, as well as the scores of all other students in the same grade. Unlike the current reporting of student performance in ELA and math, which capture achievement at a single moment in time based on grade-level standards, growth models are designed to measure how much students have learned over time.

It should be noted, the new growth scores are based on the 2016–17, 2017–18 and 2018–19 Smarter Balanced Summative Assessment results, as the state has no data from 2019–20, and the 2020–21 data was shaped by extreme variations in educational access during the pandemic. At this time, the growth model will be used for informational purposes and will not be used as part of the state’s accountability system.

On Sept. 9, the State Board of Education approved the continued use of adjusted blueprints for the Smarter Balanced assessments for the 2021–22 administration. The shortened blueprints are identical to those used in the 2020–21 academic year. Asked what effect, if any, the use of these adjusted assessment blueprints will have on the growth model, CDE Analysis, Measurement & Accountability Reporting Division Director Cindy Kazanis said during a preemptive Sept. 13 webinar the CDE is working to ensure there is as little impact as possible to the growth model.

“The State Board of Education’s action [Sept. 9] adopted the shortened blueprint for this year with plans to continue to look at this for future years,” Kazanis said. “As we develop out future years’ tests, we are working together in the accountability division and the assessments division to ensure that we have no impact to the growth model so we can produce, for the first time, a growth model with new data in 2024.”

According to Kazanis, the 2024 data release will incorporate assessment scores from 2021–22 through 2023–24.

The long road to adoption
The U.S. Department of Education pressed California to adopt a growth model when it approved the state’s plan for complying with the Every Student Succeeds Act. According to the Data Quality Campaign, California and Kansas were the only states by 2020 that didn’t have such an indicator.

After five years of engaging with stakeholders and exploring several different growth models, the CDE adopted the residual gains methodology approved by the State Board of Education in May. The State Board hired the test contractor Educational Testing Services to make technical adjustments in order to ensure the residual growth model would prove reliable and stable from year to year. The details of those technical adjustments can be found here.

The new growth model is only applied to English language arts and math test scores in grades four through eight, as prior year data will not be available for third-graders when they take their first Smarter Balanced exams, nor for students in grade 11, as they don’t take Smarter Balanced tests during their first two years of high school.

Student growth toolkit
In addition to releasing the data, the CDE also provided a toolkit on Sept. 20 to help local educational agencies and the general public understand growth scores and their appropriate uses, and how school and district leaders can effectively communicate about the model with stakeholders. The toolkit includes a fact sheet in both English and Spanish, frequently asked questions that will be updated as needed, key messages and information about appropriate uses of growth scores, as well as a brief video explaining what the growth model is and why measuring growth is important.

Any questions related to the Student Growth Model can be sent to cagrowthmodel@cde.ca.gov.

CSBA Education Policy Analyst Jeremy Anderson contributed to this article.