Data SGP is a software package used to calculate student growth percentiles and projections/trajectories from large scale, longitudinal education assessment data. The package can be used to identify students who may need extra support as well as evaluate current educational systems and identify ways to improve them. It is a more accurate measure of student progress than Value-Added Models (VAM) which only use test scores and may not capture the complexity of the learning process. This makes it an ideal candidate for future accountability systems that focus on student growth and development rather than test score achievement.
To use data sgp, you need access to long datasets of longitudinal student assessments that contain both raw scores and percentiles. The minimum requirement is sgptData_LONG which provides 8 windows (3 annually) of assessment data in LONG format from Early Literacy, Mathematics and Reading content areas. You also need a computer that has the free software environment R installed. R is a statistical programming language that can be used to perform complex calculations and analyses and has a number of higher level wrapper functions that make it easy to perform SGP analyses.
Student growth percentageiles (SGP) rank individual students’ performance relative to other students with academically similar background characteristics. Students with SGPs closer to the median show greater growth than those with SGPs below the median. SGPs are perceived as more valid and meaningful measures of student progress than simple averages of test scores because they consider a student’s past performance in addition to their current achievement level. In addition, a student’s SGP can be interpreted as a prediction of future achievement, indicating that the student is likely to achieve a certain level of success in a particular subject.
A major limitation of SGPs is that they are correlated with many individual student characteristics. For example, SGPs are correlated with prior student achievement levels, school type and teacher quality. These relationships create problems when aggregating SGPs to the teacher level. While some of this variance may be due to measurement error, a nontrivial portion of the variability in SGPs aggregated to the teacher level is likely to be explained by correlations between student characteristics and true SGPs.
Consequently, it is important to be cautious when using data sgp at the teacher level. This is especially the case when interpreting relationships between covariates and latent traits. It is also important to note that SGPs can be influenced by student background characteristics such as socioeconomic status, special education needs and student demographics. This raises the possibility that SGPs aggregated to the teacher level may not be valid indicators of teaching effectiveness. Therefore, the data sgp should be analyzed carefully before making any conclusions about teacher performance.