Student growth percentiles (SGP) are a way of reporting the relative progress a student has made on MCAS against students with similar prior achievement (their academic peers). These metrics are important because they allow us to communicate information about student achievement in terms that are intuitive and familiar to teachers and parents. They are also widely used to evaluate teacher effectiveness, in part because they provide a more unbiased and meaningful measure than unadjusted test scores. This article describes the calculation of SGPs and their pitfalls, while also exploring a more accurate alternative that avoids these problems.
The current SGP measures the extent to which a student’s performance on the most recent assessment has improved or declined relative to their academic peers. This measure is calculated using the most recent assessment and the average of two prior assessments that are matched to the same cohort. It is possible to generate a more precise estimate of a student’s SGP by conditioning on additional prior achievement attributes, but doing so may reduce the magnitude of relationships between observed student covariates and true SGPs.
A key limitation of SGPs is that they do not account for student differences in their motivation or skills, which can influence how much they grow as a result of a particular instruction. This is an important problem, but it is possible to address by estimating teacher effects using value-added models that regress student test scores on teacher fixed effects and student background variables.
The sgpData dataset provides anonymized, panel data in WIDE format that is designed to be used with the studentGrowthPercentiles and studentGrowthProjections functions. The first column, ID, provides a unique student identifier; the next five columns, GRADE_2013, GRADE_2014, GRADE_2015, GRADE_2016 and GRADE_2017, provide grade level and year values associated with the student’s assessment occurrences over 5 years. The sgpData_INSTRUCTOR_NUMBER table contains the instructor number for each student. This enables the identification of the teacher to which a student was assigned in each content area over the five years of testing. In general, the use of sgpData with the SGP functions is straightforward. However, if you are new to this data set or SGP analysis in general we recommend consulting the SGP Data Analysis Vignette for more comprehensive documentation on how to use it.