Jennifer L. Dunn Center for Assessment

12
The interaction of measurement, models and accountability: How values affect our growth model choices. Jennifer L. Dunn Center for Assessment

description

The interaction of measurement, models and accountability: How values affect our growth model choices. Jennifer L. Dunn Center for Assessment. Introduction. Can NCLB accountability be improved by growth? What do we mean by growth? - PowerPoint PPT Presentation

Transcript of Jennifer L. Dunn Center for Assessment

Page 1: Jennifer L. Dunn Center for Assessment

The interaction of measurement, models and accountability: How values affect our growth model choices.

Jennifer L. DunnCenter for Assessment

Page 2: Jennifer L. Dunn Center for Assessment

Center for Assessment2

Introduction Can NCLB accountability be improved by

growth?What do we mean by growth?

Learning that occurs beyond what the student knew when they entered the system

Movement towards proficiency

Is the construct being measured? Measurement models Accountability system One should not be considered without the other.

Page 3: Jennifer L. Dunn Center for Assessment

Center for Assessment3

Purpose Explore the properties of two NCLB growth

pilot modelsRelationship scores and classifications to

Status and Scaled score changes

Student performance in schools that are ‘helped’ by the growth model

Examine the similarities and differences of the models

Growth of proficient students, amount of required growth, and how the models value growth

Apply multiple models to the same data

Page 4: Jennifer L. Dunn Center for Assessment

Center for Assessment4

Year 1 Performance

Year 2 Performance Level

One Two Proficient Four Five

Level 1 0 200 300 300 300

Level 2 0 150 300 300 300

Proficient 0 0 300 300 300

Level 4 0 0 300 300 300

Level 5 0 0 300 300 300

NCLB Value Table

School Score = Average of value table points

Page 5: Jennifer L. Dunn Center for Assessment

Center for Assessment5

Initial Performance Growth Target

< Proficient - 1SD 1/3 of the Distance

> Proficient - 1SD &< Proficient - ½ SD

½ of the Distance

> Proficient - ½ SD &< Proficient + ¼ SD

Proficient

> Proficient + ¼ SD &< Proficient + 1¼ SD

Score – ¼ SD

> Proficient+ 1¼ SD Proficient + 1 SD

NCLB Change Model

School Score = % of students who meet or exceed target

Page 6: Jennifer L. Dunn Center for Assessment

6

Year 1 Status

0 20 40 60 80 100

05

01

50

25

0

Year 2 Status

0 20 40 60 80 100

05

01

50

25

0

Scaled Score Change

-0.4 -0.2 0.0 0.2 0.4

05

01

50

25

0

Year 1 Status

0 20 40 60 80 100

0.2

0.4

0.6

0.8

Year 2 Status

0 20 40 60 80 100

0.2

0.4

0.6

0.8

Scaled Score Change

-0.4 -0.2 0.0 0.2 0.4

0.2

0.4

0.6

0.8

NCLB Value Table NCLB Change

Page 7: Jennifer L. Dunn Center for Assessment

7

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

Sca

led

Sco

re C

ha

ng

e

Status Score

NCLB Value Table NCLB Change

PassFail

AMO = 20%

AMO = 50%

AMO = 80%

Page 8: Jennifer L. Dunn Center for Assessment

8

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

0 20 40 60 80 100

-0.4

-0.2

0.0

0.2

0.4

Sca

led

Sco

re C

ha

ng

e

Status Score

NCLB Value Table NCLB Change

PassFail

AMO = 20%

AMO = 50%

AMO = 80%

Page 9: Jennifer L. Dunn Center for Assessment

Center for Assessment9

Year 1 Performance

Year 2 Performance Level

One Two Proficient Four Five N Student

Level 1 56.7 25.6 7.3 0.4 277

Level 2 19.3 34.8 44.4 1.5 342

Proficient 2.8 8.9 70.5 14.7 3.1 1,758

Level 4 0.5 35.9 37.3 26.1 568

Level 5 7.9 24.8 67.3 355

Page 10: Jennifer L. Dunn Center for Assessment

Center for Assessment10

Initial Performance Growth TargetN

Students

% Meet

< Proficient - 1SD 1/3 of the Distance 111 46.0

> Proficient - 1SD &< Proficient - ½ SD

½ of the Distance 160 43.1

> Proficient - ½ SD &< Proficient + ¼ SD

Proficient 535 61.7

> Proficient + ¼ SD &< Proficient + 1¼ SD

Score – ¼ SD 884 87.7

> Proficient+ 1¼ SD Proficient + 1 SD 514 92.8

Total 2204 77.2

Page 11: Jennifer L. Dunn Center for Assessment

Center for Assessment11

Summary of Results

NCLB growth models:Are more related to status than scaled score

changes Relationship between scaled score change and the

NCLB change model is stronger than the VT model

“Help” more schools as the AMO increasesValue moving students toward proficient while

preventing students from falling below proficient

Page 12: Jennifer L. Dunn Center for Assessment

Center for Assessment12

Conclusion Potential to Improve NCLB accountability? It depends:

We must carefully consider: The construct, the model, the system, & the

interactions The use growth in accountability

Replace Status Multi-tier system

Adjusting the NCLB Core Principles Inclusion Different growth standards for different students Moving beyond counts of students