The data sgp package provides classes and functions for calculating student growth percentiles and projections/trajectories from large scale, longitudinal education assessment data. These calculations use a statistical methodology called quantile regression to identify the conditional density of a students achievement history and then computes the associated coefficient matrices that are used to calculate growth graphs (see image above) and percentile projections/trajectories.
To run SGP analyses you will need to have a computer running the R software program which is available for Windows, Mac OSX and Linux. We recommend a desktop or laptop that has at least 4GB of memory to ensure you have sufficient resources to run analyses without slowing down your machine. We also recommend that you familiarize yourself with the R software environment before diving in to conducting SGP analyses. There are a number of R related resources available online that can help you get started.
Student growth percentiles (SGPs) compare a students performance on the most recent MCAS test with the performances of academic peers. These academic peers are selected from all statewide students that have the same MCAS score history and can be drawn from any of the state’s demographic groups such as gender, race/ethnicity, special needs or educational programs such as sheltered English immersion or multilingual learning.
SGPs are based on a mathematical model that is expected to produce the same results for all students statewide. The model takes into account the natural fluctuations in student test scores that can occur from year to year, as well as the typical differences between students due to various factors including prior achievement and current ability level. The goal of the SGP is to provide a clear picture of how well a student is performing and how much they may need to improve in order to reach proficiency.
As you can see from the graph above, average statewide SGPs are about 50. This is not surprising since they are calculated using norms that are established using the most recent year’s worth of MCAS scores. Statewide median SGPs are not intended to be used for high stakes educator evaluation purposes and districts should begin by learning about the SGP system and exploring the BAA secure site before considering their application in educator evaluative processes.
The sgpData exemplar data set is an anonymized, panel data set comprising 5 years of annual, vertically scaled, assessment data in WIDE format. The first column, ID, provides the unique student identifier and the remaining columns provide the grades, content areas, and numeric assessment scores associated with each of these assessments over time. The sgptData_LONG data set is an additional anonymized, panel data set that models the format for the higher level wrapper functions that provide a user interface to the SGP system. The sgptData_LONG set includes a sixth column providing the student’s current grade level.
The sgptData_LONG is available from the BAA secure site under the “Reports” tab, in the “Data Files” section for Windows from the 2023/2024 School Year. When selecting the sgptData_LONG you will want to select the year(s) for which you would like to create SGP graphs and a projection/trajectory.
