Data sgp is an R package that provides classes and functions for conducting student growth percentile analyses (SGPs) using large scale, longitudinal educational assessment data. SGPs calculate relative student growth by comparing students with similar academic histories and then weighting these comparisons to derive an individual student’s rank in the cohort on the subject matter test section; higher numbers signify greater student improvement.
Educators seeking SGP information can typically register to access the state’s database and then download a report specific to their school or district that will show how students performed on each subject tested; the percentage who fell outside of the curve; and the percentage who exceeded it. Educators can use this data to inform their instructional decisions and evaluate school/district performance.
SGPs differ from traditional achievement standards in that they enable schools to establish multi-year student growth targets based on official state achievement goals or standards. This feature is especially important for communicating to stakeholders that students are progressing toward proficiency within a given time frame and enables schools to prioritize their resources for students who need it most.
To conduct an SGP, a teacher must first register to access the state’s database and download a report specific to their school or District. This report will provide them with an overview of the SGP process including a description of each step and an explanation of how the SGP is calculated. Teachers can then use the SGP report to identify areas of concern or strength and set targeted improvement goals for their students.
In order to perform an SGP, a teacher must have access to longitudinal (time dependent) student assessment data. Ideally, this data would be stored in WIDE format with each case/row representing one student and columns corresponding to variables associated with the student at different points in time. The SGP package offers the exemplar LONG format data sets sgpData_LONG and sgpData_INSTRUCTOR_NUMBER which are suitable for running SGPs (the sgpData_INSTRUCTOR_NUMBER data set also contains an anonymized lookup table that associates a teacher with each test record in its long version).
SGP calculations are based on quantitative regression analysis. To perform an SGP, a user must have a computer with the free and open source statistical software program R installed. This can be downloaded for Windows, MacOSX or Linux and requires a modest amount of familiarity with the programming language.
The SGP package provides wrapper functions abcSGP and prepareSGP that simplify the source code necessary to run these analyses. In operational SGP analyses, these functions “wrap” the lower level functions studentGrowthPercentiles and studentGrowthProjections into a single function call, reducing the number of steps necessary to run these analyses in a production setting. The following examples demonstrate the use of these wrapper functions and provide a quick introduction to SGP analyses.