Pennsylvania Value-Added Assessment System (PVAAS)
High Growth, High Achieving Schools: Is It Possible?
Fall, 2011
PVAAS Webinar
Misconception:
High achieving students/schools cannot show growth!
One Question:If PSSA is designed to discriminate proficient
from non-proficient, how can it be used to calculate growth for students at the
extremes?
Fall 2011
About the PSSA ExamsThe PSSA is designed to discriminate proficient from non-proficient, and also to have sufficient stretch to discriminate between Below Basic, Basic, Proficient and Advanced.
Other critical facts:
There is NO ceiling on the PSSA!
Each year, test scores are scaled using a lowest score of 700 and the previous year’s cut scores for the proficient category.
The high end is allowed then to be scaled based on the distribution of the data – not a fixed, pre-determined value.Fall 2011
My students are SO
high achieving already…
What is the reality of student performance on PSSA?
Fall 2011 PVAAS Statewide Core Team 6
Student NCE Scores in Mathematics
4th Grade vs. 5th Grade
Each dot represents multiple students with the same scores.
Fall 2011 PVAAS Statewide Core Team 7
Student NCE Scores in Mathematics
4th Grade vs. 5th Grade
Each dot represents multiple students with the same scores.
Fall 2011 PVAAS Statewide Core Team 8
Student NCE Scores in Mathematics
4th Grade vs. 5th Grade
Each dot represents multiple students with the same scores.
PVAAS Statewide Core Team 9
Student NCE Scores in Mathematics
4th Grade vs. 5th Grade
Notice the variation at the extremes…
Fall 2011
How does PVAAS address growth for high achieving
students and schools?
PVAAS’s Criteria for Tests
Three conditions on assessments to be included in PVAAS analyses as required by the statisticians at SAS, Inc.
1. The tests must be aligned to curriculum standards.
2. The tests must be reliable and valid.
3. The tests must demonstrate sufficient stretch to show growth of low and high performing students.
Fall 2011
PVAAS Checks for Stretch
Number of different scaled scores at the top end and at the bottom end:• Sufficient to differentiate growth at both ends
Percentages of students scoring each score at the top end and at the bottom end:• Ensures no ceilings or floors
SAS, Inc. does these checks each year!
Fall 2011
PVAAS Methodology
Multivariate Longitudinal Mixed Effect Model (MRM)
This type of model exploits the Covariance structure of the data points to dampen the error that is related to each measurement in the model.
That is, the result of the modeling process, the estimated achievement (and therefore, the estimated gain), is more precise than any of the individual measures that are used as predictors in the estimation process.
The whole is more precise than the sum of its parts!Fall 2011
So Can High Performing Groups of Students Show
Growth?
YES!!!
PVAAS Statewide Core Team 15
To Answer This Question…
Two Sources of Evidence:
1. Actual Pennsylvania Scatterplots
2. School Search of Actual PA Schools
Fall 2011
PVAAS Statewide Core Team 16
Actual Statewide 4th Grade Math Scatterplot
SY2009-10
Fall 2011
PVAAS Statewide Core Team 17
Actual Statewide 4th Grade Math Scatterplot
SY2009-10
Fall 2011
PVAAS Statewide Core Team 18
PVAAS School Search
For our purposes:
We will display actual school-wide data available on the PVAAS Public web site.
You can perform School Searches on the PVAAS secure site to allow you to view grade level results.
Fall 2011
PVAAS Statewide Core Team 19
School Search Report
Fall 2011
PVAAS Statewide Core Team 20
School Search Report
Fall 2011
PVAAS Statewide Core Team 21
School Search Report
Fall 2011
PVAAS Statewide Core Team 22
School Search Report
Fall 2011
PVAAS Statewide Core Team 23
School Search Report
Fall 2011
PVAAS Statewide Core Team 24
Now Let’s Sort The Last Column
Fall 2011
PVAAS Statewide Core Team 25
Grades 9-11 Math
Fall 2011
PVAAS Statewide Core Team 26
HIGH GROWTH A Visual Representation
Fall 2011
Compacting Simulation
Suppose we have a population of high performing students whose mean performances range from 85 to 100 (100 being the highest value).
What happens to the mean of these performances as we strive to move all students to higher performance levels?
Fall 2011
Fall 2011 PVAAS Statewide Core Team 28
Watch the Mean as the Low Increases…
80 85 90 95 100
scoresscores mean = 91.0637
low = 80
Collection 1 Dot Plot
80 85 90 95 100
scoresscores mean = 94.6823
low = 90
Collection 1 Dot Plot
80 85 90 95 100
scoresscores mean = 97.4659
low = 95
Collection 1 Dot Plot
80 85 90 95 100
scoresscores mean = 99.0533
low = 98
Collection 1 Dot Plot
PVAAS Statewide Core Team 29
YOUR SCHOOL’S DATAOpportunities for Growth?
Fall 2011
Consider Your Own Data
Investigate PSSA Data Interactive Graphical Summary: Performance
Levels Matched Comparison
Fall 2011
PSSA Graphical Summary:Performance Levels
Even very strong schools/districts have the opportunity to increase the percent of
Advanced students.Fall 2011
PSSA Matched Comparison
Even in high achieving schools, some students can slip!
Fall 2011
PVAAS Statewide Core Team 33
An Important Illustration
Fall 2011
PVAAS Statewide Core Team 34
An Important Illustration
Fall 2011
PVAAS Statewide Core Team 35
An Important Illustration
Fall 2011
36
The Report
Fall 2011
PVAAS Statewide Core Team 37
The Report
Fall 2011
In Summary…
There are high achieving schools in Pennsylvania making high growth
PVAAS provides a means to view this evidence
There are opportunities for growth in Pennsylvania’s schools
There are PA data tools to use to show where there may be opportunities for growth
Fall 2011
PVAAS Statewide Core Team 39
But, but, but…
What if my school is high achieving but not showing acceptable growth?Remember:
• Advanced and Proficient categories are ranges – students can improve or slip within those ranges… PVAAS will document those changes.
• Use School Search on the PVAAS Public or Secure site to identify and connect with schools yielding different results…
Fall 2011
Benefits of Value-Added Offers an objective, more accurate way to
measure the influence districts and schools have on students’ academic progress:
Administrators and Teachers can: Monitor the progress of all groups of students
Make more informed data-driven decisions
Align professional development efforts in the areas of greatest need
Identify best practices and implement most productive instructional strategies and programsFall 2011
PVAAS Statewide Core Team 42
www.pde.state.pa.us
Fall 2011