Data sgp is an R package that efficiently organizes longitudinal (time dependent) student assessment data to provide statistical growth plots. The data sgp package supports two common formats of this data, WIDE and LONG. While WIDE data format provides for individual case/row based analysis, LONG spreads time dependent variables across multiple rows per student. The lower level functions that do the actual SGP calculations, studentGrowthPercentiles and studentGrowthProjections, require WIDE formatted data whereas higher level wrapper functions use the LONG formatted data set. If you plan to run SGP analyses operationally year after year then we recommend using the long data set which will be easier to prepare and manage than WIDE data sets.
SGPs are calculated based on a mathematical methodology called quantile regression. This method compares a student’s current year test scores to the percentiles of students with similar prior test scores, who are known as their academic peers. This allows us to fairly compare students who enter school with different skill levels and can be used as a measure of academic progress even for students who are not yet meeting standard.
For example, if a student has an SGP of 85, that means she shows more growth than the majority of her academic peers. A student’s academic peers are all other students in Wyoming in the same grade and subject who have had comparable test scores in previous years. If a student’s current year test score is below the proficient level then she will not have an SGP. However, if her current year test score is above the proficient level then she will have an SGP.
The sgpData_WIDE data set includes assessment data from five years for each student, with the first column providing the unique student identifier and the following columns containing the scale scores associated with each year of testing. In addition, the sgpData_INSTRUCTOR_NUMBER data set is an anonymized lookup table that provides instructor number association to each test record for a given student. This is necessary because a student could be taught by multiple instructors in a single content area in a particular year.
The sgpData_LONG data set contains a similar structure to the sgpData_WIDE data, but in addition to sgpData_INSTRUCTOR_NUMBER it also contains an aggregated sgpSGPs table. This allows you to calculate sgpSGPs at any level of detail that you choose, including sgpSGPs for specific prior test scores, or for students who are not yet proficient. It also includes a teacher-instructor association lookup table which can be used to associate teachers with their student growth results. This allows for a more detailed examination of instructional practices and to detect patterns in learning styles. This can be especially useful for identifying which subjects or teachers may be having difficulties in advancing their students. The sgpData_LONG also includes an annotated version of the SGP data set that can be used for additional analyses. The annotations are a great starting point for discussing what kinds of questions would be most helpful to answer with your state team or regional consultants.