Data SGP is an online, user friendly system for calculating student growth percentiles using a longitudinal model. It uses prior achievement levels and current assessment scores to predict future performance, including how much students need to grow per year to reach an official state achievement target (also called a growth standard) for a specific grade level. SGP allows teachers and schools to communicate with parents and communities about their students’ progress by providing them with a single metric that measures both the rate of student learning as well as how far a student needs to grow to meet a specified target.
SGPs are calculated based on the historical growth trajectories of students from prior to the current school year and are updated regularly as additional testing is administered. In the Star Report, users can select a prior or current school year in the report customization section to view Window Specific SGPs for their students. These are a snapshot of the students’ current projections and may vary from the state average SGP due to the specific assessment window in which each student was assessed.
The SGPs shown in the Star Report represent estimated Student Growth Percentiles. To calculate an estimated SGP, a statistical method known as quantile regression is used to determine the relationship between a student’s previous test scores and their current assessment score. This calculation compares a student to their academic peers, or “academic neighbors.” These academic neighbors are students in the same grade and assessment subject who have similar prior test scores, i.e. who have followed a similar path of assessment score growth (Lockwood & Castellano, 2015).
These estimates are prone to errors and as such have been shown to be noisy measures of true SGPs. Furthermore, it has been found that relationships between true SGPs and student characteristics are highly variable across students. This variation, especially when aggregated at the teacher and school levels, makes it difficult to interpret relationships between true SGPs and student background.
As a result, this approach to estimating SGPs is often not able to be utilized for accountability purposes due to the difficulties in interpreting relationships between SGPs and student background characteristics at these aggregation levels. However, the SGP package offers a more accurate and reliable method for estimating student growth percentiles by directly comparing students to their academic neighbors in terms of their past achievement levels. This approach, while unable to provide predictions for all future years, is able to produce results with high accuracy and reliability for a wide variety of applications and can be used with either short or long format data sets. See the SGP package vignette for more comprehensive documentation of this analysis. The following code example shows how to use the sgpData data set with this function. The sgpData data set provides the unique student identifier, the grade level of the most recent assessment, and the scale scores of each of the 5 previous assessments. It also includes a column for sgpData_INSTRUCTOR_NUMBER, which provides an anonymized lookup table identifying the teacher with whom each student was assigned to for each of the assessments.