Data Sgp is a set of tools to facilitate educators’ access and analysis of student growth percentiles. Educators can use student growth percentiles to help identify students who may require additional support, differentiate instruction for high-achieving students and monitor the progress of all students. Growth percentiles also provide a more accurate measure of achievement than traditional mean, median or mode scores do.
The sgpData spreadsheet provides an easy-to-use, straightforward way to view and analyze student growth percentile data. The spreadsheet is formatted so educators can easily compare individual students against their peers, across grades and subjects. In addition, the sgpData spreadsheet includes student demographic data, such as gender and socioeconomic status that is not available in the BAA Secure Site summary reports.
A key feature of the sgpData spreadsheet is the ability to link student growth percentiles with teacher performance data using the teacher_id attribute in the sgpData_STUDENT_INSTRUCTORS record. This allows educators to see the actual scale score growth of a student as well as the growth percentages of all teachers who instructed that student.
To calculate a student’s SGP, the assessment system compares a current test section with at least one prior test from an earlier testing window (Fall, Winter or Spring). The results are then weighted to derive an individual’s rank within the cohort from 1 to 99; higher numbers signify improvement. To avoid spurious correlations that might result from differences in student characteristics or baseline cohort design, the SGP algorithm requires that the underlying comparisons of the current and prior tests be made with identical base populations.
The lower level SGP functions studentGrowthPercentiles and studentGrowthProjections work with WIDE formatted data. However, the higher level wrapper functions that provide a greater degree of functionality require LONG formatted data. If a district is planning on running SGP analyses operationally year after year, then we strongly recommend formatting all data sets in the long format as it will make managing the ongoing analyses much simpler.