A Computer Adaptive Alternate Assessment in Science Gary W. - - PowerPoint PPT Presentation

a computer adaptive alternate assessment in science
SMART_READER_LITE
LIVE PREVIEW

A Computer Adaptive Alternate Assessment in Science Gary W. - - PowerPoint PPT Presentation

A Computer Adaptive Alternate Assessment in Science Gary W. Phillips, Vice President & AIR Institute Fellow Paula Sable, Senior Test Developer American Institutes for Research June 22, 2016 2016 CCSSO/NCSA Outline Outline 1. Alternate


slide-1
SLIDE 1

Gary W. Phillips, Vice President & AIR Institute Fellow Paula Sable, Senior Test Developer American Institutes for Research June 22, 2016 2016 CCSSO/NCSA

A Computer Adaptive Alternate Assessment in Science

slide-2
SLIDE 2

Outline Outline

2

  • 1. Alternate Assessment State Collaboratives
  • 2. Advantages of a computer adaptive

alternate assessment

  • 3. Calibration
  • 4. Ability estimation
  • 5. Simulations for the Adaptive Algorithm
  • 6. Test Design
  • 7. Development
  • 8. Content standards – Aligned with NGSS
  • 9. Test Blueprint
slide-3
SLIDE 3
  • 1. Alternate Assessment State Collaborative
slide-4
SLIDE 4

Stat State Collaborativ e Collaboratives

 There are three collaboratives for alternate assessments based on

alternate achievement standards (AA-AAS) for students with the most significant cognitive disabilities.

1.

National Center and State Collaborative (NCSC) hosted by the National Center on Education Outcomes (NCEO) at the University of Minnesota. Development was federally funded.

2.

Dynamic Learning Maps (DLM) hosted by the Center for Educational Testing and Evaluation (CETE) at the University of

  • Kansas. Development was federally funded.

3.

Multistate Assessment Collaborative (MAAC) hosted by the American Institutes for Research (AIR) in Washington, DC involving 6 states. Development was state & AIR funded.

4

slide-5
SLIDE 5

Multistat Multistate Alt Alternat rnate Assessment Collaborativ e Assessment Collaborative

State (first Year) Grades Peer Review Approval Vertical Scale FT Designs

  • independent
  • operational
  • embedded

Growth Psychometric Model NM (2007) S Gr. 4, 7, 11 SS Gr. 11 Yes No IFT & EFT Within Grade 1 parameter Partial Credit SC (2007) S Gr. 3-5, 6-8 Bio Gr. 10 SS Gr. 3-5, 6-8 Yes No IFT & EFT Within Grades 1 parameter Partial Credit DE (2011) R,M Gr. 2, 3-5, 6-8, 9-11 S Gr. 5, 8, HS (10) SS Gr. 4, 7, HS (9) Submitted Yes R,M IFT & EFT S,SS OFT & EFT Within & Across Grades 1 parameter Partial Credit HI (2013) R,M Gr. 3-5, 6-8, 11 S Gr. 4, 8, 11 (CAT) Submitted Yes OFT & EFT Within & Across Grades 1 parameter Partial Credit OH (2013) ELA, M Gr. 3-5, 6-8, 9-11 S Gr. 5, 8, HS SS Gr. 4, 6, HS Submitted Yes OFT & EFT Within & Across Grades 1 parameter Partial Credit WY (2014) ELA, M Gr. 3-5, 6-8, 9-11 S Gr. 4, 8, 9-11 Submitted Yes OP Within & Across Grades 1 parameter Partial Credit

5

slide-6
SLIDE 6
  • 2. Advantages of an Adaptive Alternate Assessment
slide-7
SLIDE 7

Ad Adva vantages o

  • f

Adaptiv aptive Alt Alternat rnate Assessment e Assessment

Aligned to Next Generation Science Standards (NGSS)

Every student test meets the test blueprint

Higher reliability of the scores for individual students because test difficulty is adapted to student ability

Administered and scored by computer so the entire test and scoring takes about one hour

Immediate score results

Meets the same APA/AERA/NCME technical requirements as assessments of the general population

Relatively inexpensive

7

slide-8
SLIDE 8
  • 3. Item Calibration
slide-9
SLIDE 9

Calibration Calibration

9

Partial Credit Rasch Model Within-Grade Scales for grades 4, 8

and 11

slide-10
SLIDE 10

10 10

slide-11
SLIDE 11

11 11

slide-12
SLIDE 12

12 12

slide-13
SLIDE 13
  • 4. Proficiency Estimation
slide-14
SLIDE 14

Student Ability Estimation

14 14

Partial Credit Rasch Model Pattern Scoring

slide-15
SLIDE 15
  • 5. Simulations for the Adaptive

Algorithm

slide-16
SLIDE 16

Simulations Simulations

16 16

 HSA-Alt science tests are administered on an iPad as

a CAT to students in grades 4, 8, and high school.

 The simulation was based on 5000 iterations for each

test.

 Simulations were conducted to make sure that the

CAT algorithm worked as expected. Three major

  • utcomes investigated were:

% Meeting blueprint % Item exposure rate Conditional standard error of measurement (CSEM)

across the proficiency scale

slide-17
SLIDE 17

Simulation R Simulation Results sults

17 17

 No blueprint mismatch was found in the simulations.  The between student exposure rate was less than 0.2 for 43–

53% of the items and 0.4 for 57–76% of the items across the three tests. Given the limited pool size, it means that the majority

  • f the items have relatively low exposure rates.

 CSEMs between the simulation result and the 2015 science

  • perational administration.

 The red dots are from the simulation.  The plots show that the CSEMs from simulations are relatively

smaller than those from the 2015 administration, especially in the high school test. Indicates the CAT is more reliable than two-staged fixed forms (even though the pool in limited).

slide-18
SLIDE 18

CSEM CSEM

18 18

slide-19
SLIDE 19

CSEM CSEM

19 19

slide-20
SLIDE 20

CSEM CSEM

20 20

slide-21
SLIDE 21
  • 6. T
  • 6. Test Design

est Design

slide-22
SLIDE 22

Test Design f st Design for Alt

  • r Alternat

rnate CA e CAT T Assessment Assessment

 All HCPS Standards addressed at each grade (grades 4, 8, and 11 are assessed)

  • A total of 36 items is administered in grade 4.
  • A total of 40 items is administered in grade 8.
  • A total of 40 items is administered in gr. 11
  • Minimum number of items established for each standard

(varies by benchmark in each grade)

  • Maximum number of items established for each standard

(varies by benchmark in each grade)

  • Items drawn from larger pool of items within the item bank

22 22

slide-23
SLIDE 23
  • 7. Development
slide-24
SLIDE 24

De Development of Ha lopment of Hawaii AL ii ALT Assessment Standar T Assessment Standards s for Alt r Alternat rnate Assessment e Assessment

 Along with content-level educators in Hawaii, AIR/HIDOE

 Reviewed alternate assessment standards and extensions from other states  Reviewed Hawaii Content and Performance Standards  Determined the essence of each strand or domain within each standard.  Extracted key ideas from each essence to create extended standards  Content extensions have recently been revised and are now called content

specifications.

slide-25
SLIDE 25
  • 8. Content Standards
slide-26
SLIDE 26

Connecting Ha Connecting Hawaii AL ii ALT Assessment T Assessment Standar Standards t s to Ne Next Generation Science xt Generation Science Standar Standards

 AIR science staff, in conjunction with HIDOE

 Created table to note where existing HCPS intersect with NGSS  Using this bridge document, identified Alternate Assessment Content Specifications

that could be used as applications of these connections.

 Identified existing alternate assessment items that reflect these connections.  Created additional alternate assessment items to add to the pool of NGSS-related

items

26 26

slide-27
SLIDE 27

Alignment of Hawaii Content and Performance Standards to Next Generation Science Standards Examples

Grade HCPS III Strand HCPS III Standard HCPS III Benchmark description NGSS Grade/ Discipline NGSS Core Idea AIR Recommended NGSS Performance Expectation Related HCPS III Aligned Content Specifications Application 8 Physical, Earth, and Space Sciences Standard 8: Physical, Earth, and Space Sciences: EARTH and SPACE SCIENCE: Understand Earth and its processes, the solar system, and the universe and its contents. SC.8.8.2 Illustrate the rock cycle and explain how igneous, metamorphic, and sedimentary rocks are formed. Middle School Earth and Space Science Earth’s Systems MS-ESS2 Develop a model to describe the cycling of Earth’s materials and the flow of energy that drives this process. Match the type of rock to how it was formed. Identify a type of rock. Identify differences in rocks (e.g., size, texture).

27 27

slide-28
SLIDE 28

Alignment of Hawaii Content and Performance Standards to Next Generation Science Standards Examples

Grade HCPS III Strand HCPS III Standard HCPS III Benchmark description NGSS Grade/ Discipline NGSS Core Idea AIR Recommended NGSS Performance Expectation Related HCPS III Aligned Content Specifications Application 8 The Scientific Process Standard 2: Scientific Process: NATURE of SCIENCE: Understand that science, technology, and society are interrelated . SC.8.2.2 Describe how scale and mathematical models can be used to support and explain scientific data. Middle School Earth and Space Science Earth’s Systems MS-ESS2 Develop a model to describe the cycling of Earth’s materials and the flow of energy that drives this process. Identify how scale or mathematical models can be used to support and explain scientific data.

28 28

slide-29
SLIDE 29
  • 9. Test Blueprints
slide-30
SLIDE 30

Meeting the T Meeting the Test Blueprint st Blueprint

30 30

Even though each student gets a

different set of items, the items have been selected so that each set of tasks cover the state’s blueprint and content standards.

slide-31
SLIDE 31

Meeting the T Meeting the Test Blueprint st Blueprint

31 31

Standards addressed at Grades 4 and 8:

The Scientific Processes Life and Environmental Sciences Physical, Earth, and Space

Science

slide-32
SLIDE 32

Meeting the T Meeting the Test Blueprint st Blueprint

32 32

Standards addressed at High School:

 The Scientific Processes  Life and Environmental Sciences Organisms and the Environment Structure and Function of Organisms Diversity, Genetics, and Evolution

slide-33
SLIDE 33

33 33

 For each Standard, there is a block with a fixed

number of items

Example: 11 items for Gr. 11 Scientific Process  Within each block, there is a minimum and

maximum number of items per benchmark

Meeting the T Meeting the Test Blueprint st Blueprint