The data sgp package offers classes, functions and data for calculating student growth percentiles (SGP) and projections/trajectories from large scale longitudinal education assessment data. Analysis uses quantile regression models utilizing scaled score history estimates of achievement to construct coefficient matrices that generate growth percentiles that project necessary percentage progression towards reaching future achievement targets.
SGPs are calculated by comparing current student performance to that of their academic peers with similar scaled score histories from prior MCAS administrations in their subject area. Students can be compared with these peers at state, district and classroom levels to determine whether their SGPs demonstrate more, less or equivalent growth compared with their academic counterparts.
When interpreting Student Growth Profiles (SGPs), it’s important to keep in mind that percentile rankings are calculated annually, so differences between two students’ SGPs must be read with care. For instance, if Student A has higher SGP than Student B in 2023, this does not indicate greater academic growth among them; rather it could be due to certain observations being left out from calculations.
The data sgp package offers four sample data sets that can be utilized in SGP analyses. The first, sgpData, specifies data in WIDE format that’s utilized with lower level SGP functions studentGrowthPercentiles and studentGrowthProjections; second and third data sets (sgpData_LONG and sgptData_LONG) use LONG format that’s necessary for higher-level functions like abcSGP, prepareSGP and analyzeSGP; finally the fourth data set, sgpData_INSTRUCTOR_NUMBER contains an anonymized teacher-student lookup table which produces teacher level aggregates.
Running SGP analyses requires a computer running R software from either its GitHub repository or one of the download links on this page. Furthermore, SGP analyses also require some familiarity with R’s command line interface; there are plenty of resources available for learning its usage.
The SGPdata package is designed to streamline and automate these steps, making it easier for educators to conduct operational SGP analyses on their computers. Even when using this package, each step should still be followed carefully in order to produce accurate and consistent results – this is especially essential when dealing with large datasets involving calculations for all available assessments for all students in one school year. A full description of its process can be found here and should you have any queries or feedback, please reach out via the GitHub Issues Tracker as we strive to respond as quickly as possible – thanks! – SGPdata Team