DDM Implementation and Analysis

Report
DDM Part II
Analyzing the Results
Dr. Deborah Brady
Agenda





Overview of how to measure growth in 4 “common sense” ways
Quick look at “standardization”
Not all analyses are statistical or new
We’ll use familiar ways of looking at student work
Excel might help when you have a whole grade’s scores, but it is not
essential

Time for your questions; exit slips

My email [email protected];

PowerPoint and handouts at http://tinyurl.com/k23opk6
2 Considerations Local DDMs,”
1. Comparable across schools
 Example:
 Where
possible, measures are identical
 Easier
 Do

Teachers with the same job (e.g., all 5th grade teachers)
to compare identical measures
identical measures provide meaningful information about all students?
Exceptions: When might assessments not be identical?
 Different
content (different sections of Algebra I)
 Differences
in untested skills (reading and writing on math test for ELL
students)
 Other

accommodations (fewer questions to students who need more time)
NOTE: Roster Verification and Group Size will be considerations by DESE
3
2. Comparable across the District


Aligned to your curriculum (comparable content) K-12 in all disciplines

Appropriate for your students

Aligned to your district’s content

Informative, useful to teachers and administrators
“Substantial” Assessments (comparable rigor):

“Substantial” units with multiple standards and/or concepts assessed.
(DESE began talking about finals/midterms as preferable recently)
See Core Curriculum Objectives (CCOs) on DESE website if you are concerned
http://www.doe.mass.edu/edeval/ddm/example


/
Quarterly, benchmarks, mid-terms, and common end of year exams
NOTE: All of this data stays in your district. Only HML goes to DESE with a
MEPID for each educator.
Examples of
4 +1 Methods for Calculating Growth
Each is in handout
 Pre-post
test
 Repeated
 Holistic
 Post
A
measures
Rubric (Analytical Rubric)
test only
look at “standardization” with percentiles
Typical Gradebook and Distribution
Page 1 of handout

Alphabetical order (random)

Sorted low to high

Determine “cut scores” (validate in the student work)

Use “Stoplight Method” to help see cut scores

Graph of distribution of all scores

Graph of distribution of High, Moderate, Low scores
Random
90
76
92
72
80
98
91
75
60
52
76
77
96
61
63
78
79
95
80
85
86
84
65
Sorted
52
60
61
63
65
72
75
76
76
77
78
79
80
80
84
85
86
90
91
92
95
96
98
Distribution of whole class all
scores, low to high
120
100
80
60
“Cut” Scores and
“common sense”:
validate them with
performances.
What work is not
moving at an
average rate?
40
20
0
1
2
3
4
5
6
7
8
9 10 11 12 13 14 15 16 17 18 19 20 21 22 23
14
12
What work shows
accelerated
growth?
10
8
6
4
2
0
Some benchmarks
have determined
rates of growth
over time
High
Mod
Low
High, Moderate, Low
Distribution
High
Count
Mod
Count
Low
Count
6
12
5
Pre/Post Test

Description:
 The
same or similar assessments administered at
the beginning and at the end of the course or year
 Example:
Grade 10 ELA writing assessment aligned
to College and Career Readiness Standards at
beginning and end of year

Measuring Growth:
 Difference
between pre- and post-test.
 Check
if all students have an equal chance of
demonstrating growth
8
Pre- Post Tests
Pre-test
Lowest to
highest
Post test
Difference
(Growth)
Post minus pre
Analysis
Range of
growth
Pre/post
Cut score?
Look at work.
Look at distribution.
%age
growth
Difference
%age
based on
diff/pre
20
35
15
15
75%
25
30
5
5
20%
30
50
20
20
67%
35
60
25
25
42%
35
60
25
25
42%
40
70
35
35
87%
40
65
25
25
62%
50
75
25
25
50%
50
80
30
30
60%
50
85
35
35
70%
How many L/M/H?
5
3
2
LOW
MODERATE
HIGH
Holistic



Description:
 Assess growth across student work collected
throughout the year.
 Example: Tennessee Arts Growth Measure System
Measuring Growth:
 Growth Rubric (see example)
Considerations:
 Option for multifaceted performance assessments
 Rating can be challenging & time consuming
10
Holistic Example (unusual rubric)
1
No improvement in the
level of detail.
Modest improvement
in the level of detail
One is true
One is true
* No new details across
versions
* There are a few details
included across all
versions
* New details are
added, but not included
in future versions.
* A few new details are
added that are not
relevant, accurate or
meaningful
Details
2
* There are many added
details are included, but
they are not included
consistently, or none
are improved or
elaborated upon.
* There are many added
details, but several are
not relevant, accurate or
meaningful
3
4
Considerable
Improvement in the
level of detail
Outstanding
Improvement in the
level of detail
All are true
All are true
* There are many
examples of added
details across all
versions,
* On average there are
multiple details added
across every version
* At least one example
of a detail that is
improved or elaborated
in future versions
*Details are consistently
included in future
versions
* There are multiple
examples of details that
build and elaborate on
previous versions
* The added details
reflect the most relevant
and meaningful
additions
*The added details
reflect relevant and
meaningful additions
Example taken from Austin, a first grader from
Anser Charter School in Boise, Idaho. Used with
permission from Expeditionary Learning. Learn
11
more about this
and other examples at
11
http://elschools.org/student-work/butterfly-drafts
HOLISTIC
Easier for Large-Scale Assessments
like MCAS Rubric Topic or Conventions and useful when categories overlap
Criteria
In one cell
Advanced
Proficient
NI
At Risk
1)Insightful, accurate, carefully
developed claims and evidence.
2) Counterclaims are thoughtfully,
accurately, completely discussed and
argued.
3) Whole essay and each paragraph are
carefully organized and show
interrelationships among ideas.
4) Sentence structure, vocabulary, and
mechanics show control over language
use
Adequate
Effective
“Gets it”
Misconcep
tions;
some
errors
Serious
errors
Writing
1)
2)
3)
4)
Claims/evidence
Counterclaims
Organization
Language/style
MCAS Has 2 Holistic Rubrics
Topic/D
evelop
ment
Conven
tions
6
5
4
4
5
6
Rich topic/idea
development
Careful, subtle
organization
Effective rich
use of language
Full topic/idea
development
Logical
organization
Strong details
Appropriate use of
language
Moderate
topic/idea
development and
organization
Adequate,
relevant details
Some variety in
language
Rudimentary
topic/idea
development
and/or
organization
Basic supporting
details
Simplistic language
Limited or weak
topic/idea
development,
organization,
and/or details
Limited awareness
of audience and/or
task
Little topic/idea
development,
organization,
and/or details
Little or no
awareness of
audience and/or
task
Control of
sentence
structure,
grammar, usage,
and mechanics,
(length and
complexity of
essay) provide
opportunity for
student to show
control of
standard English
conventions)
Errors do not
interfere with
communication
and/or
Few errors relative
to length of essay
or complexity of
sentence structure,
grammar and
usage, and
mechanics
Errors interfere
somewhat with
communication
and/or
Too many errors
relative to the
length of the essay
or complexity of
sentence
structure,
grammar and
usage, and
mechanics
•Errors seriously
interfere with
communication
AND
•Little control of
sentence structure,
grammar and
usage, and
mechanics
Pre and Post Rubric (2 Criteria) Growth
Add the scores
Pretests
Topic
Conventio
ns
Post
tests
Topic
Convent
ions
Differe
ce
1/1
1/1
0/0
0
0
0
1 /2
2/2
1/0
1
1
100%
1/2
2/3
1/1
2
1
100%
2/3
3/3
1/0
1
2
50%
1=
2=
3=
4=
Analysis
Add together
criteria gains
as raw score
In order
% of
growth
difference
/pre
Rubrics do not represent percentages.
A student who received a 1 would probably receive a 50. F?
50
F
Seriously at risk
range 60-72, 75?
D to CAt risk
76-88, 89?
C+ to B+
Average
90-100
A to A+
Above most
Holistic Rubric or Holistic Descriptor
Keeping 1-4 scale
distribution
Pre
Post
Difference
Rank order
Cut
0
1
+1
-1
-1
0
1
+1
0
0
0
1
+1
0
0
5
1
0
-1
1
1
4
1
1
0
1
1
1
1
0
1
1
1
3
+2
1
1
1
1
0
1
1
2
3
+1
2
2
7
6
3
2
1
0
low
mod
High
Converting Rubrics to Percentages
Not recommended for classroom use because it distorts the meaning of the
descriptors.
May facilitate this large-scale use. District Decision
Pre
Converted
“grade”
Post
Converted
“grade”
Difference
%age
growth
Difference/
pre
0
0
1
50
50
50%
0
0
1
50
50
50%
0
0
1
50
50
50%
1
50
0
0
-50
-50%
1
50
1
50
0
0
1
50
1
50
0
0
1
50
3
82
32
64%
1
50
1
50
0
0
2
65
3
82
17
26%
Common Sense analysis
Was the assessment too difficult?
Zeros in pretest (3)
Zero growth
Only 1 student improved
Change assessment scale?
Look at all of the grade-level
assessments.
% conversion not helpful in this case?
Repeated Measures

Description:
 Multiple
assessments given throughout the year.
 Example:
running records, attendance, mile
run

Measuring Growth:
 Graphically
 Ranging
 Less
from the sophisticated to simple
pressure on each administration.
 Authentic
Tasks (reading aloud, running)
17
Repeated Measures



Description:
 Multiple assessments given throughout the year.
 Example: running records, attendance, mile run
Measuring Growth:
 Graphically
 Ranging from the sophisticated to simple
Considerations:
 Less pressure on each administration.
 Authentic Tasks
18
Repeated Measures Example
Running Record Errors in Reading
Average of high, moderate, and low error groups
70
Error Chart of Averages from each assessment
60
50
40
30
20
10
0
1
2
3
4
5
6
September Sept
Septe
mber
Novem Januar
March April
ber
y
June
65
48
13
30
15
15
Ra
68
63
30
35
20
22
18
10
65
65
32
22
10
12
5
2
1
30
30
28
24
22
20
19
22
Post test only
AP exam: Use as baseline to show growth for
each level or… for classroom

This assessment does not have a “normal curve”

An alternative for post test only for a classroom and to show student growth
is to give a mock AP pre and post.
Post Test Only AP Exam Example
16
14
12
10
8
6
4
2
0
five
four
three
two
one
Looking for Variability
Good
Problematic
200
150
150
# of students
# of students
200
100
50
100
50
0
0
Low
Moderate
High
Low
Moderate
High
 The second graph is problematic because it doesn’t give us information about the difference
between average and high growth because so many students fall into the “high” growth
category.
 NOTE: Look at the work and make “common sense” decisions.
21
 Consider the whole grade level; one class’s variation may be caused by teacher’s
effectiveness
 Critical Question: Do all students have equal possibility for success?
“Standardizing” Local Norms
Percentages versus Percentiles
% within class/course
%iles across all courses in district
Many Assessments
with different standards
Student A
Student A






English:
Math:
Art:
Social Studies:
Science:
Music:
15/20
22/25
116/150
6/10
70/150
35/35
“Standardized”
Normal Curve
Percentage of 100%
•
•
•
•
•
•
English
Math
Art
Social Studies
Science
Music
75%
88%
77%
60%
46%
100
Student A






English:
Math:
Art:
Social Studies:
22
Science:
Music:
62 %ile
72 %ile
59 %ile
71 %ile
70 %ile
61 %ile
Standardization
In Everyday Terms

Standardization is a process of putting different measures on the
same scale

For example
 Most
cars cost $25,000 give or take $5,000
 Most
apples costs $1.50 give or take $.50
 Getting
a $5000 discount on a car is about equal to
what discount on an apple?

Technical terms

“Most are” = mean

“Give or take” = standard deviation
23
Percentile/Standard
Deviation
Excel Functions
Sort high to low or low to high, Graphing Function, Statistical
Functions including Percentiles and Standard Deviation
 Student
grades can
be sorted from
highest to lowest
score with one
command
 Table
of student
scores can be
easily graphed
with one
command
 Excel
will easily
calculate %, but
this is probably not
necessary
“Common Sense”

The purpose of DDMs is to assess Teacher Impact

The student scores, the Low, Moderate, and High growth
rankings are totally internal

DESE (in two years) will see
 MEPIDS
 L,

and
M or H next to a MEPID
The important part of this process needs to be the focus:
 Your
discussions about student learning with colleagues
 Your
discussions about student learning with your evaluator
 An
ongoing process

similar documents