Data sgp leverages longitudinal student assessment data to create statistical growth plots (SGP) that provide visual evidence of student progress relative to academic peers. SGPs are based on latent achievement trait models that use students’ standardized test score histories and covariate information to estimate model parameters. This approach offers advantages over traditional percentile scores by reducing estimation error and providing more validity when making comparisons among students. In addition to enhancing educational decision-making, SGPs are an important tool for communicating to stakeholders that student achievement goals can be achieved within a specified timeframe. Finally, SGPs can be used as a component of educator evaluation systems by linking teachers’ performance to measurable state achievement targets/goals.
Unlike standard models, SGPs are able to make comparisons across different assessments using a common growth scale. For example, if Simon achieved a scale score of 370 on this year’s English language arts (ELA) statewide assessment, SGP will compare Simon to all other students that have taken the same assessment this school year and to all other fifth graders. SGP will also recognize that the same score on a different assessment can represent different levels of achievement, especially when the student achieves this level with a lower starting score or at a higher starting score than his/her prior performance.
SGP is a complex analysis that requires a significant amount of computing resources to produce statewide reports. Therefore, it is important that all users of the software understand the technical details involved and have a thorough understanding of the state reporting requirements. A useful reference for interpreting SGPs is the Michigan Department of Education’s Educator Reporting Guide (ERG).
In addition to a comprehensive online documentation, the sgpdata package contains an R script that allows you to generate statewide student growth plots for all students in your district. This script can be run from a command prompt or from an interactive R session. This script makes use of the SGPstateData meta-data and allows you to set a Boolean argument that indicates whether you want to anonymize student names, schools and districts in your SGP reports. The script also produces catalogs for all schools in your district that contain student growth plots. These catalogs can be downloaded in a variety of formats including PNG, PDF and JSON.
The ability to download and run these scripts is a prerequisite to utilizing data sgp. While the data sgp software provides easy to use tools for student growth analysis, it is important to understand that the underlying analysis uses advanced techniques and requires some familiarity with the programming language R. Numerous resources are available on the CRAN website to help beginners get started with R. The SGP data vignette also includes a link to an online R tutorial that is particularly helpful for newcomers to the tool. Ultimately, the success of your SGP analyses will depend on the quality and completeness of your data preparation. In most cases, errors encountered in the data analysis process revert back to issues with the data and not with the software itself.