Salient Features of India constitution especially power and functions
2011 data quality challenges when implementing classroom value added models
1. Data Quality Challenges When
Implementing Classroom Value-
Added Models
Chris Thorn
Sara Kraemer
Jeff Watson
Center for Data Quality and Systems Innovation
Value-Added Research Center
UW Madison
2. Outline
• Overview VARC/CDQSI
• Overview Value-Added methods
• Student Teacher Linkage
• Threats to validity for Classroom Value-Added
– Student teacher linkage data quality
– Assignment and matching of students and
teachers
3. VARC and CDQSI Overview
• Value-Added Research Center
Director – Rob Meyer
• Center for Data Quality and Systems Innovation
Director – Chris Thorn
• Areas of work
– Data systems / Data quality / Decision support
– Professional development
– Reporting systems
– Evaluation
– Mixed methods
• Places where we work
4. Districts and States working with VARC
Minneapolis Milwaukee
Madison Racine
Chicago New York City
US Dept. of Ed.
Los Angeles Tulsa
Atlanta
Hillsborough
County
Broward
County
6. How do we measure
student performance?
• How do we do this?
(example: the current NCLB method… percent proficient)
(example: Colorado Growth Model)
(example: VARC’s Value-Added Analysis)
• The following non-education example tries to
illustrate the difference between these measures.
8. Explaining the concept of Value-Added
by evaluating the performance of two gardeners
• For the past year, these gardeners have been tending to their oak trees trying to maximize
the height of the trees.
• Each gardener used a variety of strategies to help their own tree grow… which of these
two gardeners was more successful with their strategies?
9. To measure the performance of the gardeners, we will measure
the height of the trees today (1 year after they began tending to the trees).
• Using this method, Gardener B is the superior gardener.
This method is analogous to using an Attainment Model.
10. … but this attainment result does not tell the whole story.
• These trees are 4 years old.
• We need to find the starting height for each tree in order to more fairly evaluate each
gardener’s performance during the past year.
• The trees were much shorter last year.
Oak A Oak B
Age 4
3 Age 4
3
(1 (Today)
year ago) (1 (Today)
year ago)
11. We can compare the height of the trees one year ago to the height today.
• By finding the difference between these heights, we can determine how many inches the
trees grew during the year of gardener’s care.
• Oak B had more growth this year, so Gardener B is the superior gardener.
This is analogous to a Simple Growth Model, also called Gain.
12. … but this simple growth result does not tell the whole story either.
• We do not yet know how much of this growth was due to the strategies used by the
gardeners themselves.
• This is an “apples to oranges” comparison.
• For our oak tree example, three environmental factors we will examine are:
Rainfall, Soil Richness, and Temperature.
13. External condition Oak Tree A Oak Tree B
Rainfall amount High Low
Soil richness Low High
Temperature High Low
14. How much the gardeners’ own strategies contributed to the growth of the trees…
• We can take out each environmental factor’s contribution to growth.
• After these external factors are accounted for, we will be left with the effect of just the
gardeners.
• To find the correct adjustments, we will analyze data from all oaks in the region.
15. In order to find the impact of rainfall, soil richness, and temperature, we will plot the
growth of each individual oak in the region compared to its environmental conditions.
16. Now that we have identified growth trends for each of these environmental
factors, we need to convert them into a form usable for our calculations.
Rainfall Low Medium High
Growth in inches
relative to the -5 -2 +3
average
Soil Richness Low Medium High
Growth in inches
relative to the -3 -1 +2
average
Temperature Low Medium High
Growth in inches
relative to the +5 -3 -8
average
Now we can go back to Oak A and Oak B to adjust for their growing conditions.
17. To calculate our new adjusted growth, we start with simple growth.
• Next, we will use our numerical adjustments to account for the effect of each tree’s
environmental conditions.
• When we are done, we will have an “apples to apples” comparison of the gardeners’
influence on growth.
+14 Simple +20 Simple
Growth Growth
18. Based on data for all oak trees in the region, we found that high rainfall resulted in
3 inches of extra growth on average.
For having high rainfall, Oak A’s growth is adjusted by -3 to compensate.
Similarly, for having low rainfall, Oak B’s growth is adjusted by +5 to compensate.
+14 Simple +20 Simple
- 3 for Rainfall + 5 for Rainfall
19. For having poor soil, Oak A’s growth is adjusted by +3 to compensate.
For having rich soil, Oak B’s growth is adjusted by -2 to compensate.
+14 Simple +20 Simple
- 3 for Rainfall + 5 for Rainfall
+ 3 for Soil - 2 for Soil
20. For having high temperature, Oak A’s growth is adjusted by +8 to compensate.
For having low temperature, Oak B’s growth is adjusted by -5 to compensate.
+14 Simple +20 Simple
- 3 for Rainfall + 5 for Rainfall
+ 3 for Soil - 2 for Soil
+ 8 for Temp - 5 for Temp
21. Now that we have removed the effect of environmental conditions, our adjusted
growth result puts the gardeners on a level playing field.
We calculate that Gardener A’s effect on Oak A is +22 inches
We calculate that Gardener B’s effect on Oak B is +18 inches
+14 Simple +20 Simple
- 3 for Rainfall + 5 for Rainfall
+ 3 for Soil - 2 for Soil
+ 8 for Temp - 5 for Temp
_________ _________
+22 inches +18 inches
Adjusted Growth Adjusted Growth
22. Using this method, Gardener A is the superior gardener.
By accounting for last year’s height and environmental conditions of the trees during
this year, we found the “value” each gardener “added” to the growth of the tree.
This is analogous to a Value-Added Model.
+14 Simple +20 Simple
- 3 for Rainfall + 5 for Rainfall
+ 3 for Soil - 2 for Soil
+ 8 for Temp - 5 for Temp
_________ _________
+22 inches +18 inches
Adjusted Growth Adjusted Growth
23. How does this analogy relate to Value-Added calculations in the education context?
Oak Tree Analogy Value-Added in Education
Level of analysis? • Gardeners • IHE
• Districts/Schools
• Grades
• Classrooms
• Programs and Interventions
What are we using to • Growth in Inches • Relative Growth in Scale Score
measure success? Points
Sample • Single Oak Tree • Groups of Students
Control Factors • Rainfall • Students’ Prior Performance
• Soil Richness (most significant predictor)
• Temperature
Potential other variables collected
for ALL students
• Free/Reduced Lunch Status
• English Language Learner Status
• IEP / Special Education Status
• Race
• Gender
24. Where is Classroom-level
Value-Added Used?
• Recognizing Effective Practice
• Providing Support
• Performance Pay
• Tenure Decisions
• Higher Education Evaluation of Teacher
Preparation
• Performance Management
25. Overview of Data Requirements for
Classroom VAA
• Student demographics
• Teacher roster with background and accreditation
• School list and information
• Student and teacher attendance
• Course enrollment with assigned teachers
• Achievement data
• Mobility data
30. … and movement within schools
School
Student Teacher
Course
Course
Course
31. … and complicated workflows
School enrollment
Schedule may
is estimated
change based
on completion
School
Student Staffing needs
are estimated
Determined
sometime between
Course Feb. and July
Maybe
hired in Sept.
Teacher
32. … and absence rates
School
Student Teacher
Attendance rate, Extended leave,
discipline moved to cover,
infractions Course absences?
33. … Instructional strategies like grouping,
pull-outs, room aides, team teaching
School
Student Teacher
Instructional
Course Support staff
Specialists
35. … data errors, integration issues, SIS work
arounds, multiple districts
School SIS
Student
Student Teacher
Teacher
Matching
Records
Course HR
Data entry Creative
problems scheduling
36. Ouch
Course
Scheduling
Attendance rate,
School SIS
discipline
infractions Student Teacher Extended leave,
Student Student Teacher
moved to cover,
Matching Enrollment absences?
Records Instructional
Course HR
support staff
Data entry Creative HIRING
problems scheduling Specialists workflow
Unique School staffing
instructional adjustments
approaches
Mobility
School
37. Challenges for Classroom Level Value-
Added
• Threats to Validity
– Student teacher linkages
– Assignment of students and teachers
• Mobility
• Assessment Quality
• Assessment Coverage
• Integration
38. ST Linkages: Harvesting and Validating
Jeff Watson
Chris Thorn
Peter Witham
Timothy St. Louis
Center for Data Quality and Systems Innovation
Value-Added Research Center
UW Madison
39. 2 ST Linkage Data Quality Studies
• On-Site validation of ST Linkage data in a large
urban district
• Review of TIF districts processes for managing
and verifying ST Linkage data
40. Validation Study
Goals:
1. Determine how good / bad ST linkage data is
2. Correct bad data
3. Collect new data (e.g., team teaching)
41. Selecting a Sample
• Sample method goal is to be confident that estimates
of error rates are relatively accurate
– Sample first to assess DQ levels
• Target a small number of schools so that DQ error rates can be
estimated more accurately
• Census method goal is to confirm and adjust SIS ST
linkages to improve the validity of classroom VA
– Expanded sample as needed
• Include all schools within program
– Staged verification
42. Determine Sample Size
How to estimate sigma?
Estimate of σ (in
percent points) ±2.5% ±5.0%
10 64 16
15 144 36
20 256 64
25 400 100
30 576 144
Based on approach for determining samples size for a targeted confidence interval from
Wackerly, Mendenhall, Scheaffer (1996)
43. Target = ?? to ensure 100 Classrooms
• N=100
• Target more classrooms so that the resulting
sample in at least 100
• 75% participation rate
• 50% content rate
• Starting Sample = 100 / .75 /.5
= 267 classrooms
resulted in us recruiting about 9-10 urban
schools, mixed size and grade levels.
44. Constraints
• Control for mobility elsewhere
– Student
– Teacher
• Focus on grades and courses that are aligned
with the assessments and VA analysis
– 3-8, Math/Reading
– Formative assessments
– End of course assessments
– College readiness assessments
45. Process for Validation
• Collect data from district administration:
– Initial data files (enrollment, entity data)
– Course content case logic
• Collect data from principals:
– School – teacher assignments
– Teacher – course linkages
• Collect data from teachers:
– Teacher – course assignments
– Student-course assignments
– Team teaching
46. So…. Who is Teaching Who?
Accuracy Rate 74.2%
Team Taught Students 15.7%
Incorrect Students 2.5%
Incorrect Courses 7.6%
TOTAL 100%
47. Data from 9 schools in 1 urban district
100.0%
90.0%
80.0%
70.0%
60.0%
50.0% Accuracy Rate
40.0% Team Taught Students
30.0% Incorrect Students
20.0% Incorrect Courses
10.0%
0.0%
A B C D E F G H I
48. Counts of Student – Teacher Errors
No Add. Pull-out Push-in Targeted
Services Services Services Services NULL Grand Total
Traditional 7952 57 0 26 0 8035
Type of Classroom
Team Taught 1150 126 25 297 0 1598
“Was not in
class” 10 92 2 0 163 267
“Did not teach
class” 0 0 0 0 817 817
Grand Total 9112 275 27 323 980 10717
49. Summary of Findings
• Validated 74% of the time
• Adjusted 16% of the time
• Wrong 10% of the time
• Schools’ ST Linkage data varied between 55% and 93%
accurate
• Inaccurate records were the result of:
– not knowing about team teaching (4-30%)
– inaccurate teacher – course linkages (0-31%)
– inaccurate student – course linkages (0-15%)
50. So How Do Performance Pay Projects
Manage ST Linkage Data Quality?
1. How do TIF grantees obtain student-teacher
linkage data?
2. How do TIF grantees validate student-teacher
linkage data?
3. Are there any lessons to be learned?
51. Process Flow
Roster
Roster
Roster
Census Performance
SIS Rosters Verify Estimation
$$Payout$$
52. Process Flow
Roster
Roster
Roster
Census Performance
SIS Rosters Verify Estimation
Data
Warehouse $$Payout$$
54. Enhancing Performance Pay Programs
• Verification: Critical and Strategic
– Accuracy vetted
– Enhanced stakeholder buy-in
– Opportunity to capture nuance
• Changes in time
• Team teaching
55. Leverage Personnel for Different Purposes.
– When / where to involve district staff
– When/where to involve principals
– When/where to involve teachers
56. Organization Design
• TIF as a Business Driver (not compliance
focused; high stakes DSS)
• Multi-level connections
• Technical staff involvement
• Focus on managing teacher data
• Implementation → Challenge → Innovation
57. External partners can increase both
capacity and overhead.
• SIS vendors, Database mgmt., Verification
processes and tools
• Positive consequences:
– Increased capacity, specialized and expert
knowledge, external agency
• Negative consequences:
– Vendor silos, more project mgmt. and
communication overhead for the grantee
58. Summary
• Classroom Value-Added requires high-quality
student – teacher linkage data
• SIS Student – teacher linkage data may not
capture certain things well, like team teaching
• Verification can be staged
• Verification improves stakeholder trust and
support as well as improving the validity of VA
estimates and the decisions those estimates
inform
59. Extending the Validity and
Interpretation of
Value-Added Data
Classroom Assignment and Matching
Sara Kraemer, Robert H. Meyer, and
Robin Worth
60. Assignment and Value-Added
• To what extent are teachers systematically
assigned to certain kinds of students?
– Rothstein Critique
• What are the unique attributes of students
and teachers?
– Can we make them “observable” in the data?
• What is the process of assignment in schools?
61. Student Assignment & Classroom Value-
Added
• Classroom assignment and student-teacher
matching (validity and accuracy of VA)
– Implications for student-teacher linkages
– Sociotechnical analysis of school processes
• Sources to inform our analysis:
– Principals and former principals from 3 large
urban school districts
• 2 districts with performance compensation systems.
62. Classrooms Design and Attributes
• Purposive classroom assignment
– Heterogeneous (mix ability levels, demographics, etc…)
– Homogeneous (sort on ability levels)
– Vertical and horizontal alignment
– Ability grouping
– Split classrooms, looping, team teaching
– Other cases: Mobility, special education students
• Matching: Attributes
– Teachers
• Classroom management skills, personality, etc…
– Student
• Behavior, special cases, etc…
• Social and technical considerations
– Value-added
– Linking students to teachers
– Organizational design and process of assignment and matching
63. Value added and assignment/matching
• Extent of unbalanced classrooms?
– Ex: Teachers w/ strong classroom management
skills
• Heterogeneous assignment perceived as more
“fair”
• Implications for measuring “true” teacher
effectiveness
• Some characteristics not measured in VA
model
64. Social and organizational factors in
assignment/matching - 1
• Process of assignment and matching in schools
– Principal led
– Teams of teachers (within and across grades)
– School leader/teacher/principal hybrid team
– Collaborative versus non-collaborative model
– Timing: End of academic year
• Input factors
– Test scores, input from teachers/school leaders, principal
observation, parent input
• Environmental issues
– Teacher labor issues (turnover/retention, shortages,
teacher allocation)
65. Social and organizational factors in
assignment/matching - 2
• Contextual factors
– Levels of complexity
– Size of school, number of grades served
– Breadth and depths of courses offered
– School types
• Elementary versus high school, charters
– Demographics of student, teacher population
– Degrees of community and parental involvement
– Mobility of students across classrooms or/and schools
66. Assignment and Matching:
Linking Students to Teachers in IS
• Verification of student-teacher linkages in data
• Information systems need capture reality of
classrooms
• Assignment and matching models for
explanatory power in linkage data
67. Next steps
• How can data systems support classroom
assignment and value-added validity?
– Assessment of school assignment practices within
a district (survey development)
– System requirements for data quality validation of
student-teacher linkages
– Data systems or other tools for school-level
decision support
– Reporting to schools, includes assignment
assessment
Three different ways of measuring student performance include attainment, gain or growth, and Value-Added.The following non-education example, the Oak Tree Analogy, tries to illustrate the difference between these measures. All can be based on the same data, but the information each tells us can be very different.
This Oak Tree Analogy was created to introduce the concept of Value-Added calculations. It is not in the education context in an attempt to keep this overview of the theory of Value-Added separate from details specific to its use in education.
In this analogy, we will be explaining the concept of Value-Added by evaluating the performance of two gardeners.For the past year, these gardeners have been tending to their oak trees trying to maximize the height of the trees. Each gardener used a variety of strategies to help their own tree grow. We want to evaluate which of these two gardeners was more successful with their strategies.
To measure the performance of the gardeners, we will measure the height of the trees today, 1 year after they began tending to the trees.With a height of 61 inches for Oak Tree A and 72 inches for Oak Tree B, we find Gardener B to be the superior gardener.This method is analogous to using an Attainment Model to evaluate performance.
…but this attainment result does not tell the whole story.These gardeners did not start with acorns. The trees are 4 years old at this point in time.We need to find the starting height for each tree in order to more fairly evaluate each gardener’s performance during the past year.Looking back at our yearly record, we can see that the trees were much shorter last year.
We can compare the height of the trees one year ago to the height today.By finding the difference between these heights, we can determine how many inches the trees grew during the year of gardener’s care.By using this method, Gardener A’s tree grew 14 inches while Gardener B’s tree grew 20 inches. Oak B had more growth this year, so Gardener B is the superior gardener.This is analogous to using a Simple Growth Model, also called Gain.
But this Simple Growth result does not tell the whole story either.Although we know how many inches the trees grew during this year, we do not yet know how much of this growth was due to the strategies used by the gardeners themselves.This is an “apples to oranges” comparison.If we really want to fairly evaluate the gardeners, we need to take into account other factors that influenced the growth of the trees.For our oak tree example, three environmental factors we will examine are: Rainfall, Soil Richness, and Temperature.
Based on the data for our trees, we can see what kind of external conditions our two trees experienced during the last year.Oak Tree A was in a region with High rainfall while Oak Tree B experienced Low rainfall.Oak Tree A had low soil richness while Oak Tree B has high soil richness.Oak Tree A had high temperature while Oak Tree B had low temperature.
Our final goal is to determine how much the gardeners’ own strategies contributed to the growth of the trees.If we can determine the impact of rainfall, soil richness, and temperature on the growth of the trees, we can then take out each environmental factor’s contribution to growth.After these external factors are accounted for, we will be left with the effect of just the gardeners.To find the correct adjustments, we will analyze data from all oaks in the region.
In order to find the impact of rainfall, soil richness, and temperature, we will plot the growth of each individual oak in the region compared to its environmental conditions.On the x-axis, we plot the relative amount of each environmental condition. On the y-axis, we plot how much each tree grew from year 3 to year 4.Each dot represents a single oak tree in the area. By calculating an average line through the data, we can determine a trend for each environmental factor.From the data we collected for our region, we find that more rainfall and higher soil richness contributed positively to growth. Higher temperatures contributed negatively to growth.
Now that we have identified growth trends for each of these environmental factors, we need to convert them into a form usable for our calculations.We can summarize our trend information by determining a numerical adjustment based on High, Medium, and Low amount of each environmental condition.For example, based on our data, we found that oak trees that experienced low rainfall tended to have 5 fewer inches of growth compared to the average growth of oak trees in the region. Trees with medium rainfall tended to have 2 fewer inches of growth compared to the average. Trees with high rainfall tended to have 3 more inches of growth compared to the average.We calculate these numerical adjustments for all environmental conditions to summarize the trends from the data.Now we can go back to Oak A and Oak B to adjust for their growing conditions.
To calculate our new adjusted growth, we start with the simple growth we calculated earlier.Next, we will use our numerical adjustments to account for the effect of each tree’s environmental conditions. Essentially, we will be removing the effect of these factors.After we have removed the effect of these external conditions, we will be left with the portion of growth that was due to the gardeners’ own strategies.When we are done, we will have an “apples to apples” comparison of the gardeners’ influence on growth.
Based on data for all oak trees in the region, we found that high rainfall resulted in 3 inches of extra growth on average.For having high rainfall, Oak A’s growth is adjusted by -3 to compensate.In other words, it was easier for Oak Tree A to make growth due to rainfall conditions that were not caused by the gardener. Since we are interested in what gardener A alone contributed to the growth of the tree, we take away 3 inches of growth to remove the contribution high rainfall had on Oak Tree A’s growth.Similarly, for having low rainfall, Oak B’s growth is adjusted by +5 to compensate.It was harder for Oak Tree B to make growth due to rainfall conditions not caused by the gardener. Since we are interested in what gardener B alone contributed to the growth of the tree, we add 5 inches of growth to remove the effect of low rainfall on Oak Tree B’s growth.
We continue this process for our other environmental factors.For having poor soil, Oak A’s growth is adjusted by +3.For having rich soil, Oak B’s growth is adjusted by -2.
For having high temperature, Oak A’s growth is adjusted by +8.For having low temperature, Oak B’s growth is adjusted by -5.
Now that we have removed the effect of environmental conditions, our adjusted growth result puts the gardeners on a level playing field. What we are left with is the effect of the gardeners themselves.We calculate that Gardener A’s effect on Oak A is +22 inches.We calculate that Gardener B’s effect on Oak B is +18 inches.
Using this method, Gardener A is the superior gardener.By accounting for last year’s height and environmental conditions of the trees during this year, we have found the “value” each gardener “added” to the growth of the tree.This is analogous to a Value-Added Model.
This analogy was purposefully kept out of the education context. How does this analogy relate to Value-Added calculations in the education context?What are we evaluating? In the oak tree analogy, we evaluated gardeners. In the education context, we are evaluating districts, schools, grades, classrooms, programs, and interventions.What are we using to measure success? In the oak tree analogy, we measure growth in inches. In the education context, we measure relative growth in scale score points.What about our sample? In the oak tree analogy, we only used a single oak tree per gardener. In the education context, we use groups of students.What do we control for? In the oak tree analogy, we analyzed rainfall, soil richness, and temperature. We were then able to remove their contribution to growth. We call this “controlling” for these factors. In the education context, we control for prior performance. This tends to be the most significant predictor of student performance. Based on what other data is available for ALL students, we also control for other factors beyond the district, school, or classroom’s influence, such as:Free/Reduced Lunch StatusEnglish Language Learner StatusIEP / Special Education StatusRaceGender