Data sgp is an approach to measuring student academic achievement and growth that is more sensitive to classroom composition than value-added models. Unlike VAMs, which assign a relative ranking to students across a population of students with similar achievement histories, SGPs show the amount of growth a student has demonstrated relative to the average (or median) score of students with comparable scores on that subject-matter test. SGPs range from 1 to 99 and are interpreted like percentile ranks, with lower percentiles indicating that the student’s growth was less than the average, and higher percentiles indicating that the student’s score increased more than the average of academically-similar students.
SGP analyses rely on longitudinal student assessment data in either WIDE or LONG format and use a combination of quantitative regression methods and derived coefficient matrices to calculate student growth percentiles, projections, and trajectories. The SGP package provides exemplar data sets in both WIDE and LONG formats, sgpData and sgpData_LONG respectively, to assist in the preparation of student assessment data for SGP analysis. Because most errors associated with SGP analysis stem from incorrect or incomplete preparation of the data, it is crucial to understand the processes that are required to prepare student assessment data for SGP analyses.
DESE’s SGP methodology is designed to produce consistent, reliable and valid results. The SGP calculation process normalizes each student’s scale score (or growth measure) for all years and assessments, ensuring that the growth results are comparable across all students. The resulting SGP scores are reported to students and teachers.
Students in grades 4 through 10 have SGPs calculated for ELA and math. Students in grade 9 have SGPs calculated for science and social studies. The student-level growth graphs in the SGP dashboard display growth for all years and assessments, except the 2014-15 Badger Exam, which is omitted from this graphic because the student performance levels on the Badger exam were very different from those on the WKCE and Forward exams that preceded it.
The SGP data sets sgpData_LONG and sgpData_INSTRUCTOR_NUMBER contain teacher-student lookup tables that are used to associate teacher records with students’ test records for each content area in which they taught. A single student can have multiple teachers associated with his or her test record in a given year, and each teacher has his or her own set of SGP targets for a particular content area based on the combinations of his or her prior year SGPs for that content area.
AnalyzeSGP performs SGP analysis by using the sgpData_LONG and the sgpData_INSTRUCTOR_NUMBER datasets to construct student growth percentiles and projections, and to generate a number of other related statistics. This vignette explains how these functions are executed in the SGP package and how the outputs from these analyses are displayed in the student-level growth graphs in the SGP/Growth dashboard.
