
A student growth profile (SGP) is an indicator of students’ growth relative to their academically-similar peers. SGPs help identify trends in student achievement and progress over time, and provide an opportunity for educators to examine their instructional practices. They also can serve as a tool for detecting patterns that indicate instructional weaknesses, and help educators develop plans for correcting those weaknesses.
SGPs are created from a statistical model that regresses student test scores on teacher fixed effects, prior test scores, and student background variables. The models are estimated using a distribution compiled from multiple years of student data, so any irregularities in a single year are smoothed out. Moreover, the models include an error term that is a function of the test-specific conditional standard errors of measurement (CSEM).
While this process is computationally intensive it produces reliable estimates. The analysis combines many steps, and the bulk of the errors encountered during operational SGP analyses revert to issues with data preparation. Therefore, ensuring that data is prepared correctly is crucial to successful analyses.
Fortunately, the bulk of the work required for SGP analyses is done by lower level functions that have been written in a manner that makes them as straightforward as possible to use. This SGP package includes higher level functions that “wrap” these lower level functions, making them easier to use for operational analyses.
The lower level functions used for SGP analyses are studentGrowthPercentiles and studentGrowthProjections. Both require a large amount of memory for their operation, and they can be very computationally expensive when used with large state data sets. The higher level SGP functions abcSGP and updateSGP are designed to reduce the number of function calls needed for these operations by combining many of the steps into a single call.
In addition to these higher level SGP functions, the data sgp package contains an exemplar LONG formatted data set, sgpData_LONG, that can be used for testing and training purposes. In particular, the exemplar data set includes an anonymized instructor lookup file, sgpData_INSTRUCTOR_NUMBER, that provides insturctor information associated with each student’s test record.
The sgpData_LONG data set is available from the BAA Secure Site. To access the data, visit the “Data Downloads” page and select the “SGP Demonstration” link for either the Mathematics or ELA data set. Once the sgpData_LONG data is downloaded, open it using R and follow the instructions in the readme file to run a series of tests and training analyses on the data. The results of these analyses can then be viewed on the BAA Secure Site.