Item pool optimization for adaptive testing

Slides:



Advertisements
Similar presentations
What is a CAT?. Introduction COMPUTER ADAPTIVE TEST + performance task.
Advertisements

Smarter Balanced Assessment a Closer Look Level 1 Part 6.
Elementary Principals CCSS Update Robyn Seifert & Rita Reimbold April 10, 2013.
Welcome to Smarter Balanced Math Assessment Claims EDUCATIONAL SERVICE CENTER - NORTH LOS ANGELES UNIFIED SCHOOL DISTRICT Spring 2014 Facilitator Name.
Office of Assessment October 22, Smarter ELA/Literacy Smarter Mathematics Smarter Interim Comp Assessments Smarter Digital Library DCAS Science.
1 CSSS Large Scale Assessment Webinar Adaptive Testing in Science Kevin King (WestEd) Roy Beven (NWEA)
CALIFORNIA DEPARTMENT OF EDUCATION Tom Torlakson, State Superintendent of Public Instruction Smarter Balanced Assessment Update California Mathematics.
CALIFORNIA DEPARTMENT OF EDUCATION Tom Torlakson, State Superintendent of Public Instruction Ventura County December 2014 Interim Assessments.
Common Core State Standards: Overview of CCSS Mathematics CSU STEM Conference March 14, 2014 Ivan Cheng CSU Northridge.
Calculators Not! Why Not? Jan Martin Assessment Director, SD DOE SDCTM Feb.8, 2014.
Technical Considerations in Alignment for Computerized Adaptive Testing Liru Zhang, Delaware DOE Shudong Wang, NWEA 2014 CCSSO NCSA New Orleans, LA June.
Math Learning Progression
1 Oregon Content Standards Evaluation Project, Contract Amendment Phase: Preliminary Findings Dr. Stanley Rabinowitz WestEd November 6, 2007.
DEVELOPING ALGEBRA-READY STUDENTS FOR MIDDLE SCHOOL: EXPLORING THE IMPACT OF EARLY ALGEBRA PRINCIPAL INVESTIGATORS:Maria L. Blanton, University of Massachusetts.
Evaluating Student Growth Looking at student works samples to evaluate for both CCSS- Math Content and Standards for Mathematical Practice.
NEXT GENERATION BALANCED ASSESSMENT SYSTEMS ALIGNED TO THE CCSS Stanley Rabinowitz, Ph.D. WestEd CORE Summer Design Institute June 19,
Topics Digital Library updates
Interim Assessments Overview of Spring 2015 Assessments Training Module.
Getting Ready for MAP/EOC News from DESE & SBAC  17 states (including Missouri) voted on Nov. 14 to accept the Achievement Levels/Scale Scores.
CALIFORNIA DEPARTMENT OF EDUCATION Tom Torlakson, State Superintendent of Public Instruction Butte County Office of Education September 19, 2014 Interim.
Fall Testing Update David Abrams Assistant Commissioner for Standards, Assessment, & Reporting Middle Level Liaisons & Support Schools Network November.
Liru Zhang, Delaware DOE Shudong Wang, NWEA Presented at the 2015 NCSA Annual Conference, San Diego, CA 1.
 Closing the loop: Providing test developers with performance level descriptors so standard setters can do their job Amanda A. Wolkowitz Alpine Testing.
Idaho State Department of Education Accessing Your ISAT by Smarter Balanced Data Using the Online Reporting System (ORS) Angela Hemingway Director, Assessment.
The use of asynchronously scored items in adaptive test sessions. Marty McCall Smarter Balanced Assessment Consortium CCSSO NCSA San Diego CA.
CALIFORNIA DEPARTMENT OF EDUCATION Tom Torlakson, State Superintendent of Public Instruction Santa Clara COE Assessment Accountability Network September.
An Analysis of Three States Alignment Between Language Arts and Math Standards and Alternate Assessments Claudia Flowers Diane Browder* Lynn Ahlgrim-Delzell.
ASSOCIATION OF WASHINGTON MIDDLE LEVEL PRINCIPALS WINTER MEETING -- JANUARY 24, 2015 Leveraging the SBAC System to Support Effective Assessment Practices.
NATIONAL CONFERENCE ON STUDENT ASSESSMENT JUNE 22, 2011 ORLANDO, FL.
Practical Issues in Computerized Testing: A State Perspective Patricia Reiss, Ph.D Hawaii Department of Education.
Understanding the 2015 Smarter Balanced Assessment Results Assessment Services.
Understanding AzMERIT Results and Score Reporting An Overview.
S MARTER B ALANCED (R EQUIRED FOR DTC S, STC S, AND S MARTER B ALANCED TA S )
CTB CADDS Sally Valenzuela Director, Publishing Strategic Initiatives CTB/McGraw-Hill.
Welcome to the Interim Assessment Training for Teachers Office of Assessment March 3, 2015 While you wait for the webinar to begin, please be sure to check.
1 Oregon Standards Evaluation Project, Contract Amendment Phase: Summary of Preliminary Findings Dr. Stanley Rabinowitz WestEd December 6, 2007.
Balancing on Three Legs: The Tension Between Aligning to Standards, Predicting High-Stakes Outcomes, and Being Sensitive to Growth Julie Alonzo, Joe Nese,
California Assessment of Student Performance and Progress CAASPP Insert Your School Logo.
SBAC-Mathematics November 26, Outcomes Further understand DOK in the area of Mathematics Understand how the new SBAC assessments will measure student.
Smarter Balanced Scores & Reports. The new assessment, Smarter Balanced, replaces our previous statewide assessment, the New England Common Assessment.
Understanding the Smarter Balanced Assessment Results
What is a CAT? What is a CAT?.
ELA COE: Claim 1 Reading Information
Test Blueprints for Adaptive Assessments
PSSA Parent University
Overview of Assessments
Smarter Balanced Assessment Results
Student Growth Measurements and Accountability
What Is a Standards-Based, Computer-Adaptive Test?
Performance Task Overview
Overview of Spring 2015 Assessments
Assessment Information
Delaware Department of Education
Considerations of Content Alignment in CAT
Summative: Formative resources: Interim Assessments:
Smarter Balanced Assessment
Shasta County Curriculum Leads November 14, 2014 Mary Tribbey Senior Assessment Fellow Interim Assessments Welcome and thank you for your interest.
Aligned to Common Core State Standards
Mohamed Dirir, Norma Sinclair, and Erin Strauts
Smarter Balanced Scoring (AKA the “Marble Slides”)
Smarter Balanced Assessments
Office of Strategy, Innovation and Performance
Innovative Approaches for Examining Alignment
(Introduce new electronic score reports)
Understanding the CAASPP Student Score Reports
Presentation transcript:

Item pool optimization for adaptive testing Marty McCall Director of Psychometrics, Smarter Balanced National Conference on Student Assessment June, 2016 - Philadelphia, PA

Construct modeling and test design Content specifications derived from CCSS Assessment standards modeled on instructional standards Grade-to-grade progressions modeled Test structure follows content specs Item specifications address specific content spec elements Test blueprints derived from content specifications Depth and balance Time/cost constraints

Test Blueprint What do you want every student to get? Content – categories and proportions Cognitive characteristics Item types How many items in each test event? What are you going to report? For individuals? For groups? Overall scores Sub-scores Achievement category

Adaptive test: Item pool + test blueprint + algorithm Constrained CAT Test blueprint is a design for the student test event. It specifies the content for every student taking the test Item pool distributed to supply tests at every score level The algorithm uses the item pool to find the most informative test for each student within blueprint constraints

Pool design Reckase (2007) – P-optimal pool evaluation. “It’s unrealistic to expect that every value of theta will have a maximally informative item.” p-optimal means assuring an item within a specified proportion of full optimality for each item requested by the algorithm. Theta-scale is represented as adjacent “bins” Width of bins determined by level of optimality for desired SEM

Smarter Adaptive Test Blueprint Hierarchical relationships Claims ELA-1. Reading, 2. Writing, 3. List/Spk, 4. Research Math-1. Concepts/Procedures, 2. Problem Solving, 3. Communicating Reasoning, 4. Modeling/dData Analysis Target Clusters Targets Global rules DOK item type passage rules(ELA) Claims consistent across grades, targets consistent across grades in ELA, vary in math

Mathematics, Grade 5 Blueprint Target Specifications

Mathematics, Grade 5 Blueprint Target Specifications

Examples of non-hierarchical rules In grades 6-8, up to one CAT item per student may require hand-scoring (from either Claim 3 or Claim 4 Claim 2 (Problem Solving) and Claim 4 (Modeling and Data Analysis) have been combined .... There are still four claims, but only three claim scores will be reported with the overall math score. DOK: The CAT algorithm will be configured to ensure the following: For Claim 1, each student will receive at least 7 CAT items at DOK 2 or higher. For combined Claims 2 and 4, each student will receive at least 2 CAT items at DOK 3 or higher. For Claim 3, each student will receive at least 2 CAT items at DOK 3 or higher.

Item pool specification sheet; aka The Beast

Mathematics, grade 5 Claim 1 Beast Detail Mathematics, grade 5 Claim 1 Total N broken out by the number of bins

Bins Bin boundaries are artificial (Van der Linden method continuous, but similar) Item writers can’t predict at the fine levels required by bins They can respond to student levels Made estimates of easy, medium and hard In general, underestimated item difficulty, which has caused us to examine scale score characteristics through item mapping

Expanded pools Pools are evaluated by grade level For students at high or low extremes, the pool expands to adjacent grade levels to provide better measurement After about 2/3 of test has been administered For students who are far above or below standard Pool includes items from adjacent grades that content experts have judged to be measuring the grade level target Items with the most information are chosen from expanded pool

Pool evaluation and maintenance Uses the bin method to estimate needs Compares needs to operational pool Gives recommendations for item writing, item retirement, etc Data structure connected to the Item database for constant updating Clear and practical method

Thank you for your attention Questions? marty.mccall@smarterbalanced.org