The SGP package provides data sets for the computation of student growth percentiles (SGPs) for students in grades K-12. These data sets are referred to as the “sgpData” and can be obtained by importing the package. The sgpData data set contains state specific student assessment results for each student in the years 2013, 2014, 2015, 2016 and 2017. This article describes how to use this data to perform SGP analyses.
SGPs are a measure of student progress that combines a number of test score measures into a single index that reflects the overall academic achievement of a student. This index is then compared with the index for each student’s starting point, to determine the student’s growth rate relative to other students.
A major benefit of SGPs is that they are intended to be used as an indicator of teacher effectiveness. However, the relationships between true SGPs and student characteristics create a clear problem with interpreting aggregated SGPs at the teacher or school level.
One of the key challenges is that student characteristics can have a large effect on an individual’s growth rate. For example, if a student’s scale score on a statewide assessment increases from 300 to 370 in sixth grade, the student’s SGP will increase by 70 points. While it may be tempting to compare this growth with the growth of other students in the same grade, doing so ignores the fact that different assessments have different degrees of difficulty, and that students are at different starting points on the assessment curve. This is a source of bias in SGP estimates at the student or teacher level, and it can be avoided by using value-added models that regress student test scores on teacher fixed effects, prior test scores and student background variables.
Another source of bias comes from the fact that student performance varies across assessments and over time. This is particularly the case with students who change classrooms or schools. In addition, students who have been absent from school for significant periods of time may have lower true SGPs than their peers. The relationship between attendance and true SGPs can be modeled with covariates, but these relationships are often complicated by the fact that some covariates also have strong relationships with teacher effects.
Finally, the sgpData spreadsheet provides a convenient and straightforward way for educators to use SGP data in their instruction and evaluation of schools/districts. It is easy to use and formatted so that it is simple to compare a student’s growth from year to year for any given subject or grade. Moreover, the spreadsheet includes other information about a student, such as their gender and socioeconomic status, which is not available in the summary report. Using this data with higher-level SGP package functions requires a bit more work, but the process is relatively straightforward. Please see the SGP data analysis vignette for more comprehensive documentation on how to use this wide-format data for SGP analyses.