Data sgp is a wide format data set that stores student assessment test information in a single format that is suitable for use with the SGP package. This data set provides student growth percentiles, projections/trajectories and other education metrics. This article specifies a model for latent achievement attributes, defines true SGPs under this model and shows how their distributional properties can be assessed from the data.
Student growth percentiles (SGP) are percentage ranks of students’ current test scores relative to the average of their peers with similar prior test scores. This method of assessing student performance has gained popularity in recent years for several reasons. First, it allows comparisons across groups of students with very different backgrounds. Second, it provides a more accurate picture of student achievement than comparing unadjusted test scores. Third, it may reveal underlying causes of student achievement variation that would be difficult to uncover using traditional measures of student progress (e.g., percentages of students making expected progress on an end-of-grade test).
A SGP estimate is a noisy measure of the current value of a student’s latent achievement trait because standardized tests are error-prone and cannot accurately measure the underlying achievement characteristics of all students. These errors combine to yield estimates that are noisy measures of the student’s current latent achievement rank, which is a function of the student’s previous latent achievement rank and the current test score (Lockwood & Castellano, 2015).
Covariates may influence the SGP estimators by influencing the correlation between student background characteristics and the latent traits measured on state tests. However, the results of this study suggest that the most influential covariates explain a small percentage of the achievement variation in the SGPs, which is consistent with the hypothesis that student background influences the relationship between the student’s prior and current test scores.
The sgpData_INSTRUCTOR_NUMBER database contains the raw data used to calculate the student growth percentiles and other metrics. This data is available for download as a csv file and also as a pandas dataset. The csv file can be imported directly into the SGP toolkit to run operational analyses, while the pandas dataset can be used to explore the structural and predictive variables in the analysis.
The sgpData_INSTRUCTOR_NUMBER table is structured as follows: The first column provides the unique student identifier, ID, while the next 5 columns provide the subject and grade level for each student. The sgpData_INSTRUCTOR_NUMBER dataset also contains a set of descriptive statistics that can be used to examine the distributional properties of the student growth percentiles. Finally, the sgpData_INSTRUCTOR_NUMBER data also contains control statements that convert E-SGP group records to X-RGP records and delete the converted groups from the database. This utility makes it easy to update analyses with additional years of data without altering existing code. The higher level SGP functions require LONG formatted data and assume that the embedded sgpData_INSTRUCTOR_NUMBER meta-data has been updated to reflect the most up-to-date information for each state.