ACCESS Growth Webinar

Minnesota ACCESS Growth
Tables
Dr. Katherine Edwards
Dennis Duffy
September 2016
Outreach & Training Specialist Holly Brunson
• Provide support to both districts and
schools on testing data
• Conducting focus groups for up-coming
teacher-friendly interface
[email protected]
651-582-8563
Goals of This Presentation
• Gain an understanding of the methodology
used to calculate these growth tables
• Set a consistent method for calculating
growth using these tables
• Define appropriate and inappropriate uses
of these growth tables
ACCESS Basics
• Minnesota English language proficiency
exam since 2012
• Serves as the primary source of
information used to determine whether or
not students continue to receive EL
services
• Developed by WIDA, a non-profit
organization based out of the University of
Wisconsin
ACCESS Basics
• 4 Domains: Listening, Speaking, Reading,
and Writing
• All domains are tested separately
• The minimum proficiency score is a 1. The
maximum is a 6.
• The composite score is a weighted average
of all 4 domains
•
•
Reading and writing are weighted 35% each
Listening and speaking are each weighted 15% each
ACCESS Growth
• The Minnesota growth tables calculate
percentile growth for students in each
grade, domain, and starting proficiency
level (more on this later)
• The growth figure that you get from these
tables will report the quintile (i.e. bottom
20th percentile, 60th-80th percentile, etc.)
that students were in
ACCESS Growth vs. MCA Growth
• Unlike with MCA, there is no Z-score with
ACCESS data
• There is no set benchmark for “high
growth” or “low growth” with the ACCESS
growth. The scale score gain needed to be
in the top growth rank in 2015 may be
different than the gain needed to be in the
top growth rank in 2016
Why did Minnesota Calculate its Own
Growth Tables
• In previous years, WIDA provided growth
tables for the entire consortium and for
Minnesota
• With the introduction of ACCESS 2.0, WIDA
was unsure if it would release new growth
tables
• By calculating growth in house, it allows
MDE to ensure that growth tables are
produced using a consistent methodology
and clear parameters
Minnesota’s Tables vs. WIDA’s Tables
• Both WIDA’s growth tables and Minnesota’s
growth tables use the same underlying
methodology (percentile growth)
• Minnesota’s tables may differ from WIDA’s
depending on matching and inclusion criteria
• WIDA used a three year weighted average when
calculating the consortium-wide tables.
Minnesota’s tables do not because only one year
(2015) would have had three years of the old
version of ACCESS
How Minnesota Calculated the
Growth Tables
• Data comes from the DSR/SSR files
• Only valid scores were used
• Scores from each language domain were
examined separately
• Students were grouped by the grade they were
enrolled in for the previous year as well as by
their proficiency level from their previous year’s
score
• High school students (students enrolled in grades
9-12) were grouped together
How Minnesota Calculated the
Growth Tables
Regarding grouping students by proficiency level:
• This accounts for different growth rates at different
proficiency levels. Students at lower proficiency levels
tend to have faster raw scale score growth than students
at higher proficiency levels
• On the tables, you will see starting proficiency levels 1-1.4,
1.5-1.9, 2-2.4, 2.5-2.9, etc. up to students who started at a
level 6 (the maximum proficiency level)
• The “6” group is only for students with a previous year
proficiency score of 6
How Minnesota Calculated the
Growth Tables
• Students were matched using their MARSS ID whenever
possible. If a MARSS ID was not available, we matched
using local student IDs whenever practical
• Only students with 5 valid ACCESS domain scores
(listening, speaking, reading, writing, composite) were
included. If a student was missing a score, they were
excluded from the calculation when the tables were
created.
• Students who had multiple sets of ACCESS scores were
also excluded
How Minnesota Calculated the
Growth Tables
• For each starting proficiency level in each domain and
grade, we calculated the amount of scale score growth
from one year to the next needed to be in the bottom 20th
percentile, 20th-40th percentile, 40th-60th percentile, 60th80th percentile, and in the top 80th percentile
• Each part of the growth table represents a specific domain
and grade (e.g. listening growth for students who were in
eighth grade during the previous year).
How to Use the Growth Tables
In order to use the growth tables, you will need the following
data points:
• Individual student ID numbers (to match data from one
year to the next)
• A marker indicating which domain score you are looking at
• Previous year data on a student’s proficiency level and a
student’s previous year scale score
• The grade that a student was enrolled in during the
previous year
• A student’s current year scale score
How to Use the Growth Tables
• As an example, imagine a student named John who has
two consecutive years of ACCESS scores. In 2016, John was
in 7th grade. In 2015, he was in 6th grade. In speaking, his
2016 scale score was 350. In 2015 his speaking proficiency
score was a 2.7 and his scale score was a 320.
• We would look in the 2016 Growth Table and find the
previous year grade to be 06 and subject to be speaking
because in the previous year John was in sixth grade and
we are looking at his speaking score
How to Use the Growth Tables
• Since John had a previous year score of 2.7, we will be
looking at the numbers with a previous score range of 2.52.9
• John had a scale score gain of 30 points (350-320=30). This
gain puts John between the 40th and 60th percentile of
comparable students in speaking growth
Year
2016
2016
2016
2016
PriorYearGrade
06
06
06
06
Subject
Speaking
Speaking
Speaking
Speaking
PriorYearProficiencyLevel
2.5-2.9
2.5-2.9
2.5-2.9
2.5-2.9
Percentile
20th
40th
60th
80th
ScaleScoreChange
0
20
43
86
Important Considerations with the
Growth Tables
• There are several tables that have “ties”, or instances
where identical scale score gains will place students into
multiple growth quintiles.
• When analyzing this data in house, we have typically given
students the higher of the growth quintiles.
Year
2016
2016
2016
2016
PriorYearGrade
06
06
06
06
Subject
Speaking
Speaking
Speaking
Speaking
PriorYearProficiencyLevel
4.5-4.9
4.5-4.9
4.5-4.9
4.5-4.9
Percentile
20th
40th
60th
80th
ScaleScoreChange
-16
-1
42
42
Important Considerations with the
Growth Tables
• Any combination of domain, grade, and starting
proficiency level that had fewer than 20 students will not
have a growth rank.
• These will be filled with an asterisk
Year
2016
2016
2016
2016
PriorYearGrade
06
06
06
06
Subject
Speaking
Speaking
Speaking
Speaking
PriorYearProficiencyLevel
5.0-5.4
5.0-5.4
5.0-5.4
5.0-5.4
Percentile
20th
40th
60th
80th
ScaleScoreChange
*
*
*
*
Important Considerations with the
Growth Tables
• Even the word “growth” is at times not entirely
appropriate. Many students will get a perfect score on a
section of the test one year and lose a few scale score
points the next year.
• In these cases, it is possible to be in the top quintile by
simply not losing too many points as opposed to gaining
points.
Year
2016
2016
2016
2016
PriorYearGrade
06
06
06
06
Subject
Speaking
Speaking
Speaking
Speaking
PriorYearProficiencyLevel
6.0
6.0
6.0
6.0
Percentile
20th
40th
60th
80th
ScaleScoreChange
-59
-25
-1
-1
Appropriate Uses of the Growth
Tables
• Because of these considerations and the type of
methodology used to calculate these tables, MDE does not
recommend using these growth tables for teacher or
school accountability purposes
• We do feel that these calculations can be a useful tool in
school improvement planning. These growth quintiles can
give schools and districts a sense of where students are
progressing well or poorly relative to similar students
across the state and examine programming and resource
allocation accordingly
Distribution Tables
• The file also provides the observed distribution of students
at each proficiency level who are categorized into each
quintile of growth across the state
• These distributions are provided for each year and each
subject
• Because of ties in scale score change, there is not always
about 20% of students which fall into each quintile as
defined by the growth tables
• These are provided to allow a school or district to have a
better sense of whether they look similar in overall growth
trends to the state
Contact Info
Dennis Duffy
Division of School Support
[email protected]
651-582-8304
Katherine Edwards
Division of Statewide Testing
[email protected]
651-582-8285