Data SGP – How to Prepare and Run SGP Analyses
Data sgp leverages longitudinal student assessment data to produce statistical growth plots (SGPs) that measure students’ relative progress compared to academic peers. This analysis is often complex and time consuming and requires careful preparation to ensure accurate results. We help districts conduct these analyses by preparing the data and providing wrapper functions for running the calculations.
Using the data sgp dashboard, you can view a variety of SGP data for your students including Window Specific SGP and Current SGP. SGP is measured on a scale from 1 to 99 and higher numbers indicate greater relative growth. For example, if a student’s SGP is 75, the student has demonstrated more growth than 75% of their academic peers.
As a measure of progress, SGPs are useful for both teacher and student evaluations. SGPs are used to evaluate teachers in NJSMART’s district-level evaluation system and are a component of each educator’s overall rating score. They are also included in the federal accountability system under NCLB.
SGPs can be viewed for students in grades 4 through 11 in the Student Profile/Growth dashboard. Growth data is normalized to a single 0-100 scale and displayed for students across multiple years and assessments. These scale scores are calculated by converting the assessment-specific scale scores into SGPs using up to five years of prior test score history.
To calculate SGPs, the data sgp tool utilizes longitudinal student assessment data, student information and demographic variables. In order to run SGP analyses, you will need access to a computer that has R installed. R is a free, open source programming language that runs on a variety of operating systems. We recommend that you spend some time familiarizing yourself with R before diving into SGP analyses.
Creating a SGP from standardized testing results involves complex calculations with large estimation errors. To reduce these errors, the data sgp tools use an iterative process to prepare and analyze the data and to adjust the model parameters. To avoid errors and simplify the calculations, we recommend that you follow the step-by-step guide for preparing your data.
A number of iterations of these analyses are required to obtain a high quality SGP. To facilitate these iterations, the data sgp tool provides wrapper functions abcSGP, prepareSGP, and analyzeSGP that simplify the source code necessary to run these analyses.
The prepareSGP function takes exemplar LONG data sets, sgpData_LONG and sgptData_LONG and generates the SGP objects, Demonstration_SGP@Data, and the teacher-student lookup file, sgpData_INSTRUCTOR_NUMBER. These datasets are the input for the lower level SGP functions, studentGrowthPercentiles and studentGrowthProjections.
The analyzeSGP and combineSGP functions use the SGP object, Demonstration_SGP@Data to compute student growth percentiles and projections for each individual student in a class. These results are then merged back into the master longitudinal record, Demonstration_SGP@Data for further analysis and reporting. This process is iterative and the final result, an aggregated student growth percentile score or mSGP, will vary slightly from year to year depending on the initial model settings.