A data sgp is an estimated percentile rank of a student’s current test score relative to students with similar prior achievement levels. Data SGPs are more accurate measures of performance than standard test scores and allow educators to gain a more holistic picture of the student’s academic path.
However, these estimates are not without error. To compensate for this, SGP analyses incorporate two steps to control for estimation errors: an estimate of the student’s previous score and a comparison between the student’s current and estimated performance. As a result, these estimates are noisy and the resulting growth percentiles have large uncertainty.
In addition, the current SGP system is not flexible enough to handle a variety of student performance contexts and cannot be easily scaled to handle large datasets. This limits the application of SGP analysis to the most common scenarios.
To address these limitations, the SGP team has developed a new supplemental tool called the data sgp vignette, which is available on GitHub and provides an example of how to use the SGP package to produce student growth percentiles with a custom dataset.
The sgpData vignette is designed to allow users to generate the necessary SGP statistics in a simple, user-friendly way. It uses an exemplar panel data set that includes 5 years of vertically scaled assessment data in the WIDE format, which is the format used by the lower level SGP functions studentGrowthPercentiles and studentGrowthProjections. The vignette also contains higher level wrapper functions that help simplify the source code for operational SGP analyses.
SGPs measure progress for students at all performance levels, including those with low achievement levels. This enables educators to show students that they are improving, even if their achievement is not yet at the proficiency level. Conversely, SGPs can also provide high achieving students and schools with goals to strive for beyond proficiency.
For these reasons, it is important to understand the limitations of SGPs and how they are intended to be used. The SGP team is continuing to work towards the completion of its first goal, which is to assemble or generate multi-proxy sedimentary geochemical data (iron, carbon, sulfur, major and trace metal abundance, and trace metal isotopes) from multiple regions worldwide for every Paleozoic time slice and roughly equivalent Neoproterozoic time slices.
To run these analyses, a computer running the free and open-source R software is needed. R is compatible with Window, OSX, and Linux and, as an open-source product, can be downloaded for any operating system. It is recommended that users familiarize themselves with R before attempting to use the SGP package. Additionally, it is required that a data set be prepared in the WIDE or LONG format as the lower level SGP functions require WIDE data while the higher level wrapper functions are more flexible and utilize LONG data. Additionally, LONG data has numerous preparation and storage benefits over WIDE data.