Student Growth Percentile (SGP) Analysis in R
A student growth percentile (SGP) is a measure of a student’s relative progress or achievement based on their prior test scores. SGPs allow us to fairly compare students who enter school at different levels, and can highlight a student’s success even if they are not meeting state standards.
The SGP data sgp package provides tools for conducting normative student growth analysis using the R software environment. While the lower level SGP functions studentGrowthPercentiles and studentGrowthProjections are available for users to conduct analyses independently, the package also contains higher level function wrappers abcSGP and updateSGP that provide a streamlined SGP analysis process for use in operational settings.
In order to run SGP analyses, the data sgp package requires a computer that has the open source software R installed. This is a free and widely used statistical programming language that is available for Windows, Mac OSX, and Linux systems. If you are not familiar with the R software, we recommend spending some time to become comfortable with its basics before attempting any operational analyses.
SGP analyses rely on data sets that are in the WIDE format. The data sgpdata package includes 4 examplar data sets that can be used to demonstrate SGP analyses. One, sgpData, specifies data in the WIDE format that is used by the lower level SGP functions; studentGrowthPercentiles and studentsGrowthProjections. Two additional data sets, sgptData_LONG and sgptData_INSTRUCTOR_NUMBER, specify data in the LONG format that is used by the higher level SGP functions; prepareSGP, analyzeSGP, and summarizeSGP. sgptData_INSTRUCTOR_NUMBER contains the teacher-student lookup file utilized for producing teacher level aggregates.
When calculating a student’s SGP, the model uses up to two years of historical MCAS test performance. A student’s growth or lack thereof is compared to academic peers, which are defined as students who scored similarly on previous MCAS assessments. These academic peers do not need to be in the same subgroup as a student, such as by gender, race/ethnicity, special education status, or multilingual learning.
For each student, SGP calculations are performed to identify the range of potential trajectories that the student could follow in order to reach proficiency. These trajectory paths are then compared to the student’s current assessment to determine the Student Growth Percentile, which is reported as an individual number between 1 and 99. This SGP is then compared to the state’s target, a growth level that indicates what a student must achieve to be on track for proficiency by their end of the high school career. This target is then converted to a 1-4 scale for teacher evaluations. The mSGP is then linked to district course roster submission data and teacher practice and SGO scores for inclusion in educator evaluations.