Data sgp is software that is used to calculate student growth percentiles and/or projections/trajectories using large scale, longitudinal education assessment data (standardized tests, portfolios, grading scales or other sources). The package can be used to perform a variety of analyses including the comparison of student performance across time periods, evaluation of educational policies and practices, and examining trends over time.
Students’ progress in school is typically measured by their academic achievement or social-emotional learning (SEL). Educators use the data sgp to identify students who are struggling or making strong academic gains. The data sgp also helps educators make informed decisions about curriculum, teaching strategies and instructional supports. It is also useful for assessing teacher and leader effectiveness, especially in relation to their students’ SEL progress.
Latent achievement attributes can be identified using a variety of modeling techniques, and the distributional properties of true SGPs can be evaluated by analyzing the estimated error variance. Figure 1 shows the RMSE (relative mean squared error) as a function of the reliability l for a number of estimators of e4,2,i (conditional mean estimates). The RMSE curves are plotted separately for conditioning on different amounts of prior achievement attribute. Conditioning on additional prior achievement attribute may help to reduce relationships between unobserved student covariates and SGPs, thereby reducing the magnitude of the estimated measurement error.
The SGP package provides utility functions that support the analysis of both WIDE and LONG formatted longitudinal student assessment data. The lower level studentGrowthPercentiles and studentGrowthProjections functions operate with WIDE formatted data whereas the higher level wrapper function, sgpData, operates with LONG formatted data. In general it is recommended to use sgpData_LONG for all SGP analyses that are planned to be run operationally year after year. The sgpData_LONG vignette describes the use of this data set in detail.
A common method for aggregating estimated SGPs is to calculate them at the classroom and school level, which can be done using the SGPdata() vignette. While this allows for the identification of teacher-level variation, it is important to remember that the relationships between true SGPs and student characteristics are still present, even at an aggregated level.
Moreover, it is possible that these relationship are more important than the true SGP at the individual level. As such, there is a risk that aggregated SGPs are misleading in terms of the effectiveness of a particular program or teaching strategy. The sgpData_LONG example demonstrates this problem by showing how aggregating estimated SGPs at the school level results in a significant amount of variance that is not explained by the true student characteristics.