Data sgp is a software package that enables users to conduct student growth percentile (SGP) analyses. It is built for the R statistical software environment and assumes familiarity with using that platform. The package provides a comprehensive set of documentation, vignettes and examples for conducting SGP analyses. It also includes higher level functions such as abcSGP and updateSGP that simplify the source code required for operational analyses.
SGP is a powerful performance measure that allows schools to articulate how much students must improve over time in order to meet official state achievement targets/goals. Unlike many competing methods, SGP enables schools to establish multi-year growth standards that provide a clear picture of how far students must go to reach proficiency and help districts make a compelling case to stakeholders that improvement is necessary despite the demands of other educational priorities.
For the most accurate and precise SGP analyses, longitudinal student assessment data is required. Data must be in either WIDE or LONG format and should contain the following variables:
WIDE data format is used with the lower level SGP functions studentGrowthPercentiles and studentGrowthProjections. The higher level wrapper functions utilize the LONG data format. For all but the simplest, one-off analyses, you are likely to be better off formatting your data in the LONG format as it offers numerous preparation and storage benefits over WIDE data.
The data sgp package includes 4 examplar data sets to assist in the process of creating student assessment data for SGP analyses. The first, sgpData, is an anonymized, panel data set that consists of 5 years of annual, vertically scaled, assessment data. This exemplar data set models the format required for data used with the lower level SGP functions.
Two of the remaining data sets, sgptData_LONG and sgptData_WIDE, specify the format required for data used with higher level SGP functions such as abcSGP, prepareSGP and analyzeSGP. In addition, sgptData_LONG contains a teacher-student lookup table called sgpData_INSTRUCTOR_NUMBER that can be used to connect students to instructors by unique identifier associated with their test records.
The fourth and final data set, sgptData_LONG is an anonymized panel data set that consists of 8 windows (3 windows annually) of longitudinal student data in LONG format. The data includes the following variables: VALID_CASE, CONTENT_AREA, YEAR, ID, SCALE_SCORE, GRADE and ACHIEVEMENT_LEVEL. These variables are required for generating student aggregates by the summarizeSGP function. The rest of the variables are demographic/student categorization variables and can be optionally added if desired.