Data SGP is an analysis tool that takes longitudinal student assessment data and creates statistical growth plots (SGP) revealing how students progress relative to academic peers. SGPs take advantage of students’ standardized test score history with covariate information and use an established “growth standard” to determine growth estimates. SGPs differ from standard percentile scores because they provide a more accurate measure of student growth by adjusting for differences in starting points and comparing against academically similar students.
SGPs are a powerful new tool for educators that compare students’ performance against academically similar peers, allowing teachers and administrators to better understand how well their students are progressing and to identify areas of strength or need. SGPs are measured on a scale of 1 to 99 and can be interpreted like percentageile ranks; higher numbers indicate greater relative growth and lower numbers indicate lesser relative growth. An SGP of 75 indicates that a student’s growth was equal to or better than the growth of about half of all students with comparable score histories on the same subject-matter tests.
SGPs can be generated for individual students as well as schools, districts and states. The SGPs for a teacher are averaged to give the teacher’s mean growth percentile, or MGP, which is an indicator of how well a student performs in that teacher’s classroom. For example, an MGP of 51 means that on average students in a teacher’s class performed better than about half of their academically-similar peers. The SGPs for a school are then averaged to give the school’s overall MGP, which is an indicator of how the entire school performs in terms of student academic achievement.
The SGPs that are generated for teachers are based on the state’s existing academic achievement data which consists of students’ standardized testing results and covariate information. These are combined to generate SGPs that can be compared against the official state achievement targets/goals, which serve as a benchmark against which all teaching performance must be judged in Michigan’s educator evaluation systems.
Sgp analyses can be conducted in several ways with a number of variations in data preparation and calculation steps. In general, the lower level functions that do the actual calculations in the SGP package, studentGrowthPercentiles and studentGrowthProjections require WIDE formatted data. However, the higher level functions that serve as wrappers for the lower levels, abcSGP and updateSGP, are designed to work with LONG formatted data and can often be simplified to just one function call for operational analyses.
To conduct SGP analyses, start by downloading a set of exemplar data. The data sgpdata package contains an example WIDE format data set, sgpData, and long format data sets, sgpData_LONG and sgpData_INSTRUCTOR_NUMBER, that can be used to simulate time dependent data for use with the lower level SGP functions. The long format data sets allow you to append additional years of data without re-running the SGP functions. Managing data in long format is also much simpler than working with the WIDE formats, so we recommend using the longer data sets for operational analyses.