Data SGP is a system for producing statistical growth plots (SGP). Student growth percentiles measure how current assessment scores compare to academic peers – unlike unadjusted test scores which do not take account of prior assessments scores.
Project SGP was initiated with the intent to simplify SGP analyses as much as possible. For running calculations on SGP analyses, all that is necessary is having the R software environment installed (available for Windows, OSX and Linux and free and open source) on a computer and running calculations from there. Most time spent on SGP analyses is in prepping data before creating an analysis system for use later.
The SGPdata package offers several exemplary data sets sgpData, sgptData_LONG and sgptData_INSTRUCTOR_NUMBER that allow users to conduct SGP analyses themselves. These WIDE formatted datasets use studentGrowthPercentiles and studentGrowthProjections functions; LONG formatted data may simplify coding necessary for operational analyses.
We suggest the use of sgpstateData, a function which provides state specific meta-data for the data used in SGP analyses. You can download this information via SGPdata’s download function using sgpdata_download.
SGPs rely on quantile regression as a statistical technique, modeling the relationship between students’ current test score and those from prior exams. It is an established technique in education used for many other purposes as well.
SGP scores for students in Wyoming are determined by comparing their current assessment score against an average of academic peers that they have been academically comparable with over time. This includes all students in their same grade and assessment subject who have taken similar Star assessments over time, matching them up using factors such as past SGP performance, current standardized test score comparisons and an algorithm to predict aggregated past performance on Star assessments.
Understand that student growth percentiles (SGPs) are estimates of student performance and therefore may contain some estimation errors. Also keep in mind that two students with different scale scores could share the same SGP.
Contrasting large community databases like Genbank and EarthChem, Data sgp focuses on research questions directly relevant to participants involved in the project. This offers clear incentive for participation as it will allow researchers access to vast amounts of metadata and legacy data that might otherwise remain unavailable to researchers. Eventually, our aim is to migrate this data into full community databases; however, given its immense size this may take time; hence our efforts at quickly providing low cost access and analyses on it in the meantime.