MDE will offer both SGP and Value-Added Models for student growth considerations
|
There are states who have successfully implemented different growth models through continuous stakeholder involvement and assuring the evaluation process focuses on teacher engagement and effective feedback. There are 18 Race to the Top (RTT) who were given millions of dollars to implement student growth into educator evaluations, among other requirements for the federal grant. Most of these states have moved away from a "simple or gain score model" to primarily one of three alternatives: Value-Added Models (VAM), Residual Gain Model, or Student Growth Percentiles (SGP), learn more about SGP from the video on the right from Georgia or Virginia.
|
|
|
Exemplary states such as Colorado balance student growth and student achievement. The student profile on the left shows how an individual student achieved over a three year period by placing a white dot at one of four levels: Developing, Approaching, Meets, Exceeds. The colored arrows describe how the student grew (SGP) relative to others across the state: Low growth (red), Average (white), High growth (green). Finally, there is a projection band for the next year depending on if the student demonstrates high, average, or low growth.
|
Understanding Student Growth Percentiles from BAA
Although the BAA Secure Site released SGPs for individual students (4th - 11th grade) in January, 2016, we concur with the strong recommendation of MDE that SGPs should NOT be used for educator evaluations in 2015/16. In fact, the current law (PA-173) does not require the use of SGPs until 2018/19 which allows three more years to stabilize state-level data. BAA has released the SGPs in order for educators to familiarize themselves with the data prior to high-stakes use in 2018/19.
|
A few key points to understand, SGPs across the state are NOT a normal distribution (bell-shape curve), in fact, the distribution is expected to have an equal number of students at each percentile (the diagram on the right shows grouping of 10 or deciles). In this example, there are 100 students in 5th, 7th and 11th grade; note that each decile (grouping of 10 percentiles) has exactly 10 students. Granted, this is the expected value and what is seen across the state with a large number of students. A school with fewer students would be expected to have similar results, though grade levels may vary. As a general rule of thumb, the average SGP will be 50 (or between 40-60), the median SGP will be 50 (or between 40-60) as well and the distribution will be fairly flat.
There are two other examples to the right that show a unique situation where the majority of the SGPs are in the higher percentiles (sloped upward from L to R) and the median is 57. This is what schools would hope to see, however, others may see a downward slope where the majority of SGPs are in the lower percentiles. This represents a group that did not show much growth from one year to the next on the state level tests a median SGP of only 37. 
|
Achievement and GrowthIn addition to the SGP Tool created by Clare-Gladwin, many educators are accustomed to seeing growth paired with student achievement. Macomb ISD has created a macro driven MS Excel document that will take the same district student data file from BAA Secure Site and plot both achievment and growth.
Featured on the left is the 2016 M-STEP results for 8th grade math for a fairly large, sample data set, suppose it is a large district. The color bands represent the four proficiency levels of the M-STEP (Not Proficient, Partially Proficient, Proficient or Advanced). For the SAT data, there are only two levels set to the college readiness benchmark (do not meet or met). The vertical axis running through the Partially Proficient region represents the state average achievement score. The horizontal axis is set at the 50th percentile for Student Growth Percentiles. NOTE: With both the M-STEP and SAT results seen on the left, there seems to be a relationship between achievement and growth. The lower achievement tends to have lower growth and the higher achievement tends to have higher growth. This likely holds true for statewide data as well. |
Understanding the Growth and Achievment Tools
To utilize the interactive data tools created by Macomb and Clare-Gladwin ISDs, you must download the "Student Data File" under "Reports" on the BAA Secure Site, then copy/paste DATA ONLY (not including formulas or header) into one of the data tools. The podcasts below will help walk you through the process and explain a little bit about the tool. At this point, districts should begin learning about SGPs, not applying them to educator evals.
1. Download from BAA3. Copy/Paste Student Data (Values ONLY)5. Understanding Achievement & Growth Quadrants
|
2. Downloading Tools
4. Understanding SGP Distributions (bar graphs)6. Filtering Data and Cautions with SGPs
|
SGP Tools, compliments of Macomb ISD and Clare-Gladwin RESD:
- M-STEP Growth and Achievement (includes the bar graphs and the quadrant graphs with filters)
- SAT Growth and Achievement (includes the bar graphs and the quadrant graphs with filters)
