Beyond 3-8 Reading and mathfinal_gg

Report
Moving Beyond Math &
Reading 3-8: How Value-Added
Models Can Incorporate Other
Types of Assessments
Garron Gianopulos, PhD
North Carolina Department of Public Instruction
Marianne Motley,
Ohio Department of Education
John White, PhD
SAS, Instititute
June 25, 2014
Moving Beyond Math & Reading 3-8: How Value-Added
Models Can Incorporate Other Types of Assessments
Marianne Mottley, Assistant Director, Office of Accountability
Ohio Department of Education
“A principal or a superintendent must be
able to say to the school board and the
public:
Everyone who teaches here is good – and
here’s how I know.”
- Charlotte Danielson
Ultimate goal of
teacher
evaluation
Improve
teaching and
learning
How do we get there?
Evaluation Framework
Teacher
Performance
on Standards
50%
Student
Growth
Measures
50%
Final Summative Rating
Accomplished
Skilled
Developing
Ineffective
RttT Teacher-Student Linkage
Spring
2011
Spring
2012
Spring
2013
30%
of Ohio’s
LEAs
Participated
60%
of Ohio’s
LEAs
Participated
100%
of Ohio’s
LEAs
Participated
Linkage
Required Linkage Grades 4-8
- Reading
- Math
Optional Linkage for Extended Testing
- Grades K-3 reading and math
- Grades K-4, 6, 7 science
- Grades K-8 social studies
- Grades 9-12 all content areas
Extended Testing Programs
RttT Mini-Grant
- Grades 1-3 R/M
- Grades 2-3 Science
- Grade 3 Social Studies
Project SOAR
- Grade 3 R/M
- Grades 3-8 Social Studies
- Grades 3, 4, 6, 7 Science
- High School EOC Exams
RttT Mini-Grant
Not a mandatory part of RttT
Districts could opt in by submitting a minigrant application
District agreement with ODE
Two-Year Grant = Two-Year Commitment
RttT Mini-Grant Overview
Grant pays for assessments in grades K-3
and for teacher value added reports
Pools created based on which
assessments were most widely used
Three pools in Ohio
- Terra Nova (CTB McGraw Hill)
- MAP (NWEA)
- STAR (Renaissance Learning)
Mini-Grant Overview
Univariate Response Model (URM)
Predicted Mean Approach
Tests Properties
- High correlation to curricular objectives
- Have sufficient stretch
- Must be sufficiently reliable
Mini-Grant Overview
Univariate Response Model (URM)
Minimum 3 prior year’s or years’ tests
(predictor tests)
Pool must have same predictors
Pool must have same current year tests
(response tests)
RttT Mini-Grant Overview
Predictor Testing Requirements
Terra Nova
- Spring window in at least 3 subjects
- Reading and math mandatory
- Science or social studies optional
MAP/STAR
- Fall & spring windows in at least 2 subjects
- Reading and math mandatory
Mini-Grant Overview
Some districts had prior test data and
received reports in 2013
For others, 2014 will be their first year
when reports will be generated
Pools must remain stable across years
Districts cannot deviate from a pool
Mini-Grant Overview
Reports must be used for teacher
evaluation
Considered to be a vendor assessment for
the evaluation system
Must be used for at least 10% of the 50%
that comprises student growth measures
Project SOAR
Battelle for Kids Project SOAR
Created in 2002 with 42 districts
2014 over 100 SOAR districts
Uses value-added data to drive
decision making and enhance student
learning
Must use for teacher evaluations
Project SOAR Grades 3-8
Predictor Data
Response Data
Grade 2 Terra Nova
R/M/S/SS
Grade 3 reading & math
using Ohio achievement
assessments
All available prior years’
data
Grades 3-8 social studies
using Terra Nova social
studies test
All available prior
years’ data
Grades 3, 4, 6, 7 science
using Terra Nova
science test
Project SOAR High School
Uses ACT Quality Core
End-of-Course Exams
Four Content Areas
Mathematics
English/Language Arts
Algebra I
Geometry
Algebra II
Pre-Calculus
English 9
English 10
English 11
English 12
Science
Biology
Chemistry I
Social Studies
U.S. History
Other BFK Clients
NWEA and possibly other vendors
Using “single year” Multivariate Response
Model (MRM) for Year 1
- Uses fall tests as baseline data
- Uses spring tests to measure growth
Using “across years” MRM approach for
Year 2 and beyond
Other BFK Clients
Districts will partner directly with BFK
Exact grades/subjects yet to be determined
Grade 3 reading and math will use Ohio’s
achievement assessments as response
(using URM calculation)
education.ohio.gov
Social Media
Ohio Families and Education
Ohio Teachers’ Homeroom
ohio-department-of-education
storify.com/ohioEdDept
@OHEducation
OhioEdDept
Moving Beyond Math &
Reading 3-8: How Value-Added
Models Can Incorporate Other
Types of Assessments
Garron Gianopulos, PhD
Psychometrician, Technical Development of
Statewide Assessments, Design and
Implementation of Tests,
North Carolina Department of Public Instruction
June 25, 2014
Overview
 A Year of Change
 How NC Uses EVAAS
 Outcomes and Plans for the Future
2012-13: A Year of Change
READY Assessments
• Introduced all new READY end-of-grade (EOG)
Assessments in grades 3 through 8
• New READY end-of-course (EOC) assessments
• All EOGs and EOCs were aligned to the common
core state standards,
adopted
2010
2005: Implemented
new growth in
formula
• All new standards classifying students as ready for
the next grade level and on-track to college
• New proficiency levels
28
2012-13: A Year of Change
New Laws
• Read to Achieve
• Law required third graders to demonstrate
proficiency in reading
• 35 new third grade portfolio reading assessments
2005: Implemented3
new
growth formula
• Beginning-of-grade
Reading
assessment
• School Letter Grades
• Grade letter determined by proficiency and EVAAS
growth
29
2012-13: A Year of Change
Race to the Top Assessments for Teacher Evaluation
• New Teacher Effectiveness model required teachers to
demonstrate growth
• 6 New Occupational Course of Study (OCS)
assessments
• 30 New NC Final
Examsnew growth formula
2005: Implemented
• Science exams in grade 4,6,7, and high school
• New H.S. Math exams
• Social Studies exams in grade 4 – 9 and high school
• Historically non-tested subjects
30
2012-13: A Year of Change
State Board of Education (SBE)
• Not all board members were comfortable with the new
proficiency levels
• A policy decision was made to change original 4
achievement levels to 5 achievement levels
• The new level
differentiates
between
readiness for next
2005:
Implemented new growth
formula
grade level and being on-track for college
• NC is a governing member of the Smarter Balanced
Assessment Consortium (SBAC); however, the SBE is
re-evaluating consortia and future testing options
• Keeping current READY assessments through 2015-16
31
Overview
 A Year of Change
 How NC Uses EVAAS
 Outcomes and Plans for the Future
How NC Uses the Educator Value
Added Assessment System (EVAAS)
• Educator Instructional Planning
• School Accountability
• Educator Effectiveness
33
Instructional Planning
 2012-13: EVAAS provides projections of
EOG/EOC percentile scores at the
beginning of the school year to help with
instructional plans
 Student-level EVAAS projections are also
provided for ACT, PLAN, and EXPLORE
 2013-14: EOG/EOC Scale score
projections will be provided
The School Accountability System
35
North Carolina Educator Evaluation Process
Teachers
We have a total of 6 standards in our teacher evaluation
system. All standards, 1-6, are of equal value. Our goal:
• Identify our strongest teachers and explore their
methodologies, and
• Support teachers who need to increase their
effectiveness
• Three years of EVAAS growth measures needed before
ratings apply
1 2 3 4 5 6
Demonstrate
Leadership
Establish
Environment
Know
Content
Facilitate
Learning
Reflect on
Practice
Contribute
to
Academic
Success
Training Provided
Webinars with EVAAS team focused on how staff
will continue to get value-added data with a
change in standards
–Understanding the EVAAS system of valueadded growth
–Conversion of assessment data to Normal
Curve Equivalents (NCE’s)
–Understanding the use of assessments in
3
various growth models
7
EVAAS Models Used in NC
Multivariate Response Model (MRM)
• 17 End-of-grade Math and ELA assessments
• Used in the school accountability system
Univariate Response Model (URM)
• 3 End-of-course assessments
• 6 Occupational Course of Study assessments
• 70 Career and Technical Exams
• 30 NC Final Exams (historically non-tested subjects)
38
• All available assessments used in educator
effectiveness
What Types of Scores Can be Used in
EVAAS?
•
•
•
•
•
Sufficient spread in the scale
Limited ceiling or floor effects
Correlation between covariates and score > .60
High reliability
Test content needs to align closely to the
instructed curriculum
• High Validity
• Standardized assessments need curricula that
39
are also standardized
Overview
 A Year of Change
 How NC Uses EVAAS
 Outcomes and Plans for the Future
Outcomes from 2012-13
READY EOG/EOC Assessments
• Public backlash against quantity of testing (especially in
grade 3), but not against our chosen growth model
• All READY EOG/EOC assessments and most Final
Exams had the required properties to be used by EVAAS
• No changes were needed to the READY EOG/EOC
assessments as these were developed over a three year
time frame with sufficient funding
• The legislatively-driven calculations for school letter
grades in the Accountability Model generated a
41
disproportionately large percent of Fs and Ds
Outcomes from 2012-13
Final Exams
• Certain Final Exams and OCS assessments did not
meet requirements to be utilized in the EVAAS
models
– Insufficient sample size
– Insufficient correlations between covariates and
test score
– Insufficient reliabilities
• Floor effects surfaced on some High School Math
assessments
42
Outcomes from 2012-13
Constructed Response Items (CR)
• NCDP provided general guidelines for managing
CR item scoring process, but each district
implemented their own process
• Constructed response (CR) items too labor
intensive to score by teachers
• Many concerns about the security, inter-rater
reliability, and accuracy of teacher-scored CR items
43
Outcomes from 2012-13
Response from Educators
• The large majority (79%) of teachers met or
exceeded growth expectations
• Teachers with insufficient individual estimates were
excluded from the rating system
• Few groups are comparing their own growth
measures to EVAAS
• Most educators seem to be responding favorably to
the EVAAS reports
44
Changes in 2013-14
Managerial Changes
• New chief in charge of the test development section
• The test development section is now managing the test
development process for all Final Exams
• Increased test development staff to manage work load
45
Changes in 2013-14
Final Exams
• Name was changed from Common Exams to Final
Exams to emphasize that they are intended to replace
teacher-made final exams
• Testing time increased from 90 minutes to 120 minutes
• Reduced the number of Final Exams with CR items
• Reduced the number of CR items per exam
• Tests lengthened with MC items to increase reliability
• CR items scored by contractor rather than by NC
46
teachers
Changes in 2013-14
Final Exams
• Moving many of the Final Exams to online
administration
• Implementing common item non-equivalent equating
design
• Experimental items embedded to provide greater
statistical control in form assembly
• Reduced difficulty of some Math High School tests to
improve reliability and remove floor effects
47
Changes in 2013-14
• Providing more detail in test specifications to help
teachers know what to instruct in those subjects where
the curriculum standards are not specific enough
• The EVAAS team will determine if they can use the
beginning-of-grade 3 assessment to measure growth
48
2013-14: More Change Ahead?
• The weight given to EVAAS growth within the school
Accountability Model will likely increase to produce a
more acceptable distribution of letter grades
• Two bills approved by senate and the house to rescind
the Common2005:
Core
State Standards
Implemented new growth formula
• Will funding for Final Exams remain at the same level?
49
Every Student READY
Questions

similar documents