Value-Added Model
|
Value-added isa statistical model that uses student-level growth scores to differentiate teacher performance in the area of student learning growth. It is the state-adopted method for interpreting FCAT reading and math scores for use in teacher evaluations. |
Why value-added?
The Student Success Act requires the inclusion of student learning growth measures in teacher evaluations. That act also tasks the education commissioner with identifying and implementing student learning growth models. It also states,
“Beginning in the 2011-2012 school year, each school district shall measure student learning growth using the formula approved by the commissioner under paragraph (a) for courses associated with the FCAT.” -1012.34(7)(b), F.S.
The American Institute for Research was hired to analyze Florida's FCAT data and to provide information to the state's Student Growth Implementation Committee about statistical models that would provide suitable options for using FCAT scores as a measure of teacher effectiveness. After reviewing the research and evaluating the data, the committee selected a covariate adjustment value-added model.
How is value-added different?
A value-added model measures the impact of a teacher on student learning by accounting for other factors that may impact the learning process. These statistically significant factors, or covariates, include:
Up to two prior years of achievement scores
-
Number of subject-relevant courses taken
-
Disability status
-
English language learner status
-
Gifted status
-
Mobility
-
Attendance
-
Difference from modal age
-
Class size
-
Homogeneity of prior test scores
These covariates are run through the statistical model to establish a predicted, or expected performance level for each student. The model is substantially different than other models that may evaluate teachers based on a single year of student performance or evaluate teachers based on simple comparisons of student growth from one year to the next.
How is a student's predicted, or expected, performance level determined?
Prior test scores are the strongest predictor for student growth. Prior achievement data is combined with other measured student characteristics (such as the covariates above) to form the predicted, or expected, performance of the student.
Specifically, the formula is this:
In any case, the point is that prior test scores are statistically the greatest predictor of future student performance. That data is added to the other student characteristics to determine the predicted performance as illustrated below.
The key is to remember the predicted performance is a combined representation of prior performance, along with other measurable characteristics about the student and environment; such as attendance, age, disability status, class size, the number of subject-relevant courses the student is taking, etc. Some of these factors effectively lower the student’s expected performance level, while others may raise it.
How does a student's predicted or expected performance fit into our teacher evaluation system?
A student's predicted performance serves as the target. A student whose actual performance meets or exceeds predicted performance counts as a positive benefit to the teacher in terms of the teacher's evaluation as illustrated in the following chart.
This student's achievement is higher than was predicted. The teacher's evaluation will be higher as a result of this student's performance.
---------------------------------------------------------------------------------
A student whose actual performance does not meet or exceed his predicted performance counts as a negative attribute to his teacher in terms of the teacher's evaluation as illustrated below.
This student's achievement is lower than was predicted. The teacher's evaluation will be lower as a result of this student's performance.
How is a teacher's effectiveness rating determined from the students' achievement as measured by the value-added model?
The percent of students whose actual performance was equal to or higher than their predicted performance is calculated for each teacher. Those percentages are then compared to the scale below for classification purposes:
| Student Learning Growth Rating: | Range (%) |
|---|---|
| Highly Effective | |
| Effective | |
| Needs Improvement / Developing | |
| Unsatisfactory |
*Remember that the numbers shown in the "range" column of the chart represent the percent of students who achieved at or above their expected performance level based on their statistical value-added calculation.
This student growth calculation counts as half of the overall evaluation.