Data sgp is an important part of the software package for SGP analyses. It contains the classes, functions and data that are used to calculate student growth percentiles and percentile growth projections/trajectories using large scale, longitudinal education assessment data. Data sgp also provides information on the distributional properties of these statistics.
The vignettes contained in this section describe the use of WIDE and LONG data formats for SGP analyses. For most analyses, it is recommended that you format the data in the LONG format since it offers numerous preparation and storage benefits over the WIDE format. This is particularly true if you plan to run your SGP analyses operationally year after year.
This vignette illustrates how to construct a student growth percentile (SGP) from a single prior test score for students in grades 4 through 8 and in grade 10. The SGPs calculated by DESE compare students’ current tests scores with their prior tests scores. For example, the SGP for a grade 10 student compares the student’s current English language arts (ELA) and math tests scores with their prior tests scores in those two subjects.
In general, SGPs estimated from standardized test scores are error-prone measures of their corresponding latent achievement attributes. These errors are due to finite sample sizes, biases in item selection and construction, as well as other model and methodological issues (Akram, Erickson & Meyer, 2013; Lockwood & Castellano, 2015; McCaffrey, Castellano, & Lockwood, 2015).
To estimate the accuracy of SGPs estimated from standardized test scores, it is necessary to understand their distributional properties. For this purpose, this vignette describes the distribution of true SGPs for individual students, their correlations with each other, and their relationship to student background characteristics.
The vignette also demonstrates how to improve the quality of SGPs by conditioning on additional information about the student. The plot shows RMSEs of SGPs conditional on only the prior ELA test scores (curve with triangles), on only the prior math tests scores (curve with X) and on both the math scores and the covariates (curve with +).
In addition to the vignettes, this documentation contains several examples of SGP analyses that you can use to practice your skills and gain proficiency in the SGP software package. If you have a specific SGP question that is not answered in this documentation, feel free to ask it in the SGP discussion forum. We will be happy to help you! SGP is an open source project. You can contribute your code, bug fixes and suggestions by submitting an issue on the GitHub repository. Alternatively, you can contact the SGP developers directly through email at [email protected]. You can download and use the SGP software for free, but be sure to cite the authors when you distribute your work. This software is released under the BSD License. For a complete explanation of the license and other terms and conditions, see the license document.