New Meridian State Leads Meeting Growth

New Meridian State Leads · February 2, 2021

New Meridian State Leads Meeting Growth — New Meridian State Leads, February 2, 2021.
Published

February 2, 2021

Event
Invited talk
Slides
18
Format
Converted from PowerPoint

This deck is an image-based copy of the original slides. Builds and animations are flattened; the slide text is reproduced below for search and screen readers.

Slide text
  1. Proposed SGP Analyses for 2020-2021 Damian Betebenner & Adam Van Iwaarden State Leads Meeting February 2nd, 2021 This work is licensed under a Creative Commons Attribution 4.0 International License.
  2. A Brief History of Consortium Growth • Growth analyses were first conducted for the consortium in Fall 2015 and Spring 2016. • In 2015-2016 there were 9 states/organizations a part of the consortium. • Growth analyses have been conducted in ELA and Mathematics including several math end-of-course content areas (e.g., Algebra I, Integrated Math) • Growth analyses have been conducted each Fall and Spring up until Fall 2019. Proposed SGP Analyses for 2020-2021 2
  3. A Brief History of Consortium Growth • In calculating growth, we have attempted to give states/ organizations the maximum amount of SGP-related data we can derive including: • State/organization and consortium referenced SGPs • State/organizations and consortium referenced measurement error corrected SGPs (SGP SIMEX) • Growth to standard student growth projections/SGP targets (catch-up/ keep-up and move-up/stay-up) using both state/organization and consortium norms. Proposed SGP Analyses for 2020-2021 3
  4. A Brief History of Consortium Growth • Analyses have not been without challenges • Mode of administration (online versus paper-pencil) effects (beginning 2017). • Changing state membership altering the meaning of a consortium norm (ongoing). • ABO versus Flagship forms (2019) • Pre- and post-equating occurring in the same year for different states/ organizations (2019) • SGPs are a magnifying glass: If there are issues with the underlying assessments, growth will allow one to see them (i.e., growth can be used to show problems with growth results). • We want to help Pearson make the assessments as sound as possible as it will make the growth data derived from the tests as sound as possible. Proposed SGP Analyses for 2020-2021 4
  5. Recap of 2019-2020 Testing • Fall 2019 testing occurred. Growth was for these data in February 2020. • As we all know, the pandemic began in March 2020. • Except for about 80,000 students in Illinois who tested before the onset of the pandemic, spring 2020 testing didn’t occur and was cancelled. • The Feds gave accountability and testing waivers for the 2019-2020 academic year. Proposed SGP Analyses for 2020-2021 5
  6. Status of 2020-2021 Testing • The pandemic raged on in Fall 2020. • Fall 2020 testing was cancelled. • Currently (as of early February 2021) it appears as though most states intend to try and administer spring 2021 state assessments (there are a few states that have applied for testing waivers). • Regarding 2020-2021 accountability, states have been able to submit addenda or waiver requests to their plans. Thus far, there is no blanket cancellation of accountability. • In what follows we presume that testing occurs in some form in spring Proposed SGP Analyses for 2020-2021 6
  7. Hopes for 2020-2021 Testing • Due to the absence of 2020 data SGPs will be calculated from 2019 to 2021 (skip-year growth) and utilize up to two priors (2018 and 2019). • Our goal is to provide states/organizations with all the growth quantities they have always received (plus some additional stuff) • State/organization and consortium referenced SGPs • State/organizations and consortium referenced measurement error corrected SGPs (SGP SIMEX) • Student growth projections/Growth to standard SGP targets (catch-up/ keep-up and move-up/stay-up) using both state/organization and consortium norms. (Growth targets will require using historical sequential- year growth norms in order to perform projections forward from 2021) • We also hope to calculate baseline referenced SGPs using historical skip- year norms. (More on this later) Proposed SGP Analyses for 2020-2021 7
  8. Challenges to Calculating SGPs in 2021 • We expect numerous challenges to conduct the analyses as we have laid out. • Secure forms versus remote forms (could yield issues like we saw with P&P versus online) • Different platforms being used to administer the test (PA5 versus PAN) • Participation rates • Motivation effects • Baseline referenced analyses assume that the scale used currently has not drifted from the one from which the baseline coefficient matrices are derived. • Challenges impact analyses at different levels: • At the test level: both scaled score comparability within and across years. • Secure vs remote is a within-year comparability issue. Baseline referenced analyses if an across year issue. • Participation rates likely have the largest impact when aggregating results to the group level Proposed SGP Analyses for 2020-2021 8
  9. Challenges to Calculating SGPs in 2021 • We are approaching these issues in two ways: • Issues that we can simulate (e.g., low participation rates) we are running sensitivity analyses and simulations on to examine the potential impact and consider what correctives are available. • Issues that we have difficulty simulating (e.g., secure versus remote forms) we are planning to run analyses immediately after the arrival of the data to understand impact. • We will also have backup plans to conduct analyses depending upon the outcome. For example, secure versus remote administration. Proposed SGP Analyses for 2020-2021 9
  10. Hopes for 2020-2021 Testing (In addition) • The pandemic is a once in a century occurrence that has disrupted education nationwide. • We believe that state summative assessments have the potential to provide significant insight into the impact of the pandemic on student learning (often referred to as Learning Loss) • To assist states/organizations in examining learning loss we intend to additionally calculate baseline referenced SGPs (both state and consortium referenced) • Baseline referencing will utilize 2016, 2017 to 2019 skip-year data (using data from states currently in the consortia or a previous set of states???). • Baseline coefficient matrices are constructed that are subsequently used with 2018, 2019 to 2021 skip-year data. • The mean SGP for the baseline years is 50. Presumably, the pandemic has significantly impacted student learning and we expect mean baseline SGPs to be lower in 2021. Proposed SGP Analyses for 2020-2021 10
  11. Illustration of Learning Loss Proposed SGP Analyses for 2020-2021 11
  12. Illustration of Magnitude of Learning Loss Proposed SGP Analyses for 2020-2021 12
  13. Using Growth to Monitor Learning Loss Precautions associated with learning loss interpretation • Use of baseline SGP imply that the 2021 results are on the same scale as the 2019 results. Pearson will work hard to ensure that. • Illinois 2020 to 2021 growth results for the 80,000 or so students will help validate learning loss results for 2019 to 2021. • Decreased participation rates can cause issues when aggregating to the groups level • Will leave “holes” in the data for many districts and schools. • Holes in the data will have less an impact on growth results than on status results because growth is uncorrelated with prior achievement. • At present, we think learning loss for those districts/schools with “holes” can be estimated using whatever 2021 student growth data is available together with some technique like propensity score matching at the district/school level. • Again, this isn’t meant to be used for “big A” accountability Proposed SGP Analyses for 2020-2021 13
  14. Using Growth to Monitor Learning Loss Current efforts • Our goal (working with Pearson) is to have all preliminary work on learning loss (e.g., baseline matrix construction) done prior to the arrival of data (preliminary data usually arrives in June). • Create learning loss reports for each state in addition to traditional SGP reports. • Turn around reports to states/organizastions as fast as possible so that data is actionable for Fall 2021. Proposed SGP Analyses for 2020-2021 14
  15. Using Growth to Monitor Learning Loss Future efforts • The pandemic’s effects will not disappear after this spring. • Ideally, this spring’s assessments will allow us to understand the impact of the pandemic on student learning. • This information will be critical to help inform policies designed to ameliorate the impact. • Equally (and perhaps even more) critical will be monitoring the recovery of students going forward in 2021-2022, 202— Proposed SGP Analyses for 2020-2021 15
  16. Learning Loss & Learning Recovery Proposed SGP Analyses for 2020-2021 16
  17. Monitoring Learning Recovery • Monitoring recovery from the impact of the pandemic will be critical. • State summative assessment data can provide critical insight into ongoing recovery. • Widening achievement gaps can occur due to both differential impact with the pandemic as well as differential recovery. • Beware the K-shaped recovery where advantaged students predominantly bounce back and disadvantaged students predominantly don’t bounce back. Proposed SGP Analyses for 2020-2021 17
  18. Questions Proposed SGP Analyses for 2020-2021 18